Daniel Miessler's Blog, page 10
June 10, 2024
A Collection of Thoughts and Predictions About AI (June 2024)
Member Post
I have been talking about AI and making limited predictions about it for nearly 10 years now, but some new ideas and thoughts have come out recently that I want to comment and expand on.
I'm specifically speaking of the wondrous conversation between Leopold Aschennbrenner and Dwarkesh Patel, and Leopold's collection of essays he recently released on the topic of AI safety (my summary here).
That’s what kicked off a bunch of ideas, and I now have so many I need to make a list.
Table of ContentsHow I see the AI paths of development
Reacting to Leopold and Dwarkesh's conversation
Thoughts on the ladder to AGI and then to ASI
So that’s what I’m going to cover in this long post, done in the jumping style of my annual Frontview Mirror pieces that I do for members.
How I see the ladder path to ASI / Consciousness
Subscribe to Unsupervised Learning Membership to read the rest.Become a paying subscriber of Unsupervised Learning Membership to get access to this post and other subscriber-only content.
Upgrade Translation missing: en.app.shared.conjuction.or Sign In
A subscription gets you: Access to the UL community and chat (the thinking and sharing zone) Exclusive UL member content (tutorials, private tool demos, etc.) Exclusive UL member events (currently two a month) More coming!Powered by beehiiv
June 5, 2024
Podcast Summary: Dwarkesh vs. Leopold Aschenbrenner
This is a Fabric conversation extraction (using the extract_wisdom_dm pattern) of the 4-hour conversation between Dwarkesh and Leopold about AGI and other topics.
SUMMARYLeopold Aschenbrenner discusses AGI timelines, geopolitical implications, and the importance of a US-led democratic coalition in developing AGI.
IDEAS- CCP will attempt to infiltrate American AI labs with billions of dollars and thousands of people.
- CCP will try to outbuild the US in AI capabilities, leveraging their industrial capacity.
- 2023 was when AGI went from a theoretical concept to a tangible, visible trajectory.
- Most of the world, even those in AI labs, do not truly feel the imminence of AGI.
- The trillion dollar AI cluster will require 100 GW, over 20% of current US electricity production.
- It is crucial that the core AGI infrastructure is built in the US, not authoritarian states.
- If China steals AGI weights and seizes compute, it could gain a decisive, irreversible advantage.
- AGI will likely automate AI research itself, leading to an intelligence explosion within 1-2 years.
- An intelligence explosion could compress a century of technological progress into less than a decade.
- Protecting AGI secrets and infrastructure may require nuclear deterrence and retaliation against attacks.
- Privatized AGI development risks leaking secrets to China and a feverish, unstable arms race.
- A government-led democratic coalition is needed to maintain security and alignment during AGI development.
- Solving AI alignment becomes more difficult during a rapid intelligence explosion with architectural changes.
- Automated AI researchers can be used to help solve AI alignment challenges during the transition.
- AGI may initially be narrow before expanding to transform robotics, biology, and manufacturing.
- The CCP's closed nature makes it difficult to assess their AGI progress and strategic thinking.
- Immigrant entrepreneurs like Dwarkesh Patel demonstrate the importance of US immigration reform for progress.
- Trillions of dollars are at stake in the sequence of bets on the path to AGI this decade.
- Many smart people underestimate the intensity of state-level espionage in the AGI domain.
- Stealing the weights of an AGI system could allow an adversary to instantly replicate its capabilities.
- Algorithmic breakthroughs are currently kept secret but could be worth hundreds of billions if leaked.
- Small initial advantages in AGI development could snowball into an overwhelming strategic advantage.
- AGI may be developed by a small group of top AI researchers, similar to the Manhattan Project.
- Privatized AGI development incentivizes racing ahead without caution in order to gain a market advantage.
- Government-led AGI development can establish international coalitions and domestic checks and balances.
INSIGHTS- The US must proactively secure its AGI development to prevent a catastrophic strategic disadvantage.
- Leaking of AGI algorithms or weights to adversaries could be an existential threat to liberal democracy.
- Policymakers and the public are unprepared for the speed, scale, and stakes of imminent AGI progress.
- Privatized AGI development is incompatible with the coordination and caution required for safe deployment.
- A government-led international coalition of democracies is essential to maintain control over AGI technology.
- Immigration reform to retain top foreign talent is a critical strategic priority for US AGI leadership.
- Scenario planning and situational awareness are crucial for navigating the complex path to AGI.
- Hardening AGI labs against state-level espionage will require military-grade security beyond private capabilities.
- Timely and decisive government intervention is needed to nationalize AGI before a private lab deploys it.
- Humanity must proactively shape AGI to respect democratic values, rule of law, and individual liberty.
QUOTES- "The CCP is going to have an all-out effort to like infiltrate American AI labs, billions of dollars, thousands of people."
- "I see it, I feel it, I can see the cluster where it's stained on like the rough combination of algorithms, the people, like how it's happening."
- "At some point during the intelligence explosion they're going to be able to figure out robotics."
- "A couple years of lead could be utterly decisive in say like military competition."
- "Basically compress kind of like a century worth of technological progress into less than a decade."
- "We're going to need the government to protect the data centers with like the threat of nuclear retaliation."
- "The alternative is you like overturn a 500-year civilizational achievement of the government having the biggest guns."
- "The CCP will also get more AGI pilled and at some point we're going to face kind of the full force of the ministry of State security."
- "I think the trillion dollar cluster is going to be planned before the AGI, it's going to take a while and it needs to be much more intense."
- "The US bared over 60% of GDP in World War II. I think the sort of much more was on the line. That was just the sort of like that happened all the time."
- "The possibilities for dictatorship with superintelligence are sort of even crazier. Imagine you have a perfectly loyal military and security force."
- "If we don't work with the UAE or with these Middle Eastern countries, they're just going to go to China."
- "At some point several years ago OpenAI leadership had sort of laid out a plan to fund and sell AGI by starting a bidding war between the governments."
- "I think the American National Security State thinks very seriously about stuff like this. They think very seriously about competition with China."
- "I think the issue with AGI and superintelligence is the explosiveness of it. If you have an intelligence explosion, if you're able to go from kind of AGI to superintelligence, if that superintelligence is decisive, there is going to be such an enormous incentiveto kind of race ahead to break out."
- "The trillion dollar cluster, 100 GW, over 20% of US electricity production, 100 million H100 equivalents."
- "If you look at Gulf War I, Western Coalition forces had 100 to 1 kill ratio and that was like they had better sensors on their tanks."
- "Superintelligence applied to sort of broad fields of R&D and then the sort of industrial explosion as well, you have the robots, you're just making lots of material, I think that could compress a century worth of technological progress into less than a decade."
- "If the US doesn't work with them, they'll go to China. It's kind of surprising to me that they're willing to sell AGI to the Chinese and Russian governments."
- "I think people really underrate the secrets. The half an order of magnitude a year just by default sort of algorithmic progress, that's huge."
- "If China can't steal that, then they're stuck. If they can't steal it, they're off to the races."
- "The US leading on nukes and then sort of like building this new world order, that was kind of US-led or at least sort of like a few great powers and a non-proliferation regime for nukes, a partnership and a deal, that worked. It worked and it could have gone so much worse."
- "I think the issue here is people are thinking of this as chat GPT, big tech product clusters, but I think the clusters being planned now, three to five years out, may well be the AGI superintelligence clusters."
- "I think the American checks and balances have held for over 200 years and through crazy technological revolutions."
- "I think the government actually like has decades of experience and like actually really cares about this stuff. They deal with nukes, they deal with really powerful technology."
- "I think the thing I understand, and I think in some sense is reasonable, is like I think I ruffled some feathers at OpenAI and I think I was probably kind of annoying at times."
- "I think there's a real scenario where we just stagnate because we've been running this tailwind of just li
ke it's really easy to bootstrap and you just do unsupervised learning next token prediction."
- "I think the data wall is actually sort of underrated. I think there's like a real scenario where we just stagnate."
- "I think the interesting question is like this time a year from now, is there a model that is able to think for like a few thousand tokens coherently, cohesively, identically."
HABITS- Proactively identify and mitigate existential risks from emerging technologies like artificial intelligence.
- Cultivate a strong sense of duty and responsibility to one's nation and the future of humanity.
- Develop a nuanced understanding of geopolitical dynamics and great power competition in the 21st century.
- Continuously update one's worldview based on new evidence, even if it contradicts previous public statements.
- Foster international cooperation among democracies to maintain a strategic advantage in critical technologies.
- Advocate for government policies that promote national security and protect against foreign espionage.
- Build strong relationships with influential decision-makers to shape the trajectory of transformative technologies.
- Maintain a long-term perspective on the societal implications of one's work in science and technology.
- Cultivate the mental flexibility to quickly adapt to paradigm shifts and disruptive technological change.
- Proactively identify knowledge gaps and blindspots in one's understanding of complex global issues.
- Develop a rigorous understanding of the technical details of artificial intelligence and its potential.
- Seek out constructive criticism and dissenting opinions to pressure-test one's beliefs and assumptions.
- Build a strong professional network across academia, industry, and government to stay informed.
- Communicate complex ideas in a clear and compelling manner to educate and influence public discourse.
- Maintain a sense of urgency and bias towards action when confronting existential risks to humanity.
- Develop a deep appreciation for the fragility of liberal democracy and the need to defend it.
- Cultivate the courage to speak truth to power, even at great personal and professional risk.
- Maintain strong information security practices to safeguard sensitive data from foreign adversaries.
- Proactively identify and mitigate risks in complex systems before they lead to catastrophic failures.
- Develop a nuanced understanding of the interplay between technology, economics, and political power.
FACTS- The CCP has a dedicated Ministry of State Security focused on infiltrating foreign organizations.
- The US defense budget has seen significant fiscal tightening over the past decade, creating vulnerabilities.
- China has a significant lead over the US in shipbuilding capacity, with 200 times more production.
- AGI development will likely require trillion-dollar investments in compute and specialized chips.
- The largest AI training runs today use around 10 MW of power, or 25,000 A100 GPUs.
- Scaling AI training runs by half an order of magnitude per year would require 100 GW by 2030.
- The US electric grid has barely grown in capacity for decades, while China has rapidly expanded.
- Nvidia's data center revenue has grown from a few billion to $20-25 billion per quarter due to AI.
- The US produced over 10% of GDP worth of liberty bonds to finance World War II spending.
- The UK, France, and Germany all borrowed over 100% of GDP to finance World War I.
- The late 2020s are seen as a period of maximum risk for a Chinese invasion of Taiwan.
- China has achieved 30% annual GDP growth during peak years, an unprecedented level in history.
- AlphaGo used 1920 CPUs and 280 GPUs to defeat the world's best Go player in 2016.
- The Megatron-Turing NLG has 530 billion parameters and was trained on 15 datasets.
- The number of researchers globally has increased by 10-100x compared to 100 years ago.
- The US defense budget in the late 1930s, prior to WWII, was less than 2% of GDP.
- The Soviet Union built the Tsar Bomba, a 50 megaton hydrogen bomb, in the 1960s.
- The Apollo program cost over $250 billion in inflation-adjusted dollars to land humans on the Moon.
- The International Space Station required over $100 billion in multinational funding to construct.
- The Human Genome Project cost $3 billion and took 13 years to sequence the first human genome.
REFERENCES- The Chip War by Chris Miller
- Freedom's Forge by Arthur Herman
- The Making of the Atomic Bomb by Richard Rhodes
- The Gulag Archipelago by Aleksandr Solzhenitsyn
- Inside the Aquarium by Viktor Suvorov
- The Idea Factory by Jon Gertner
- The Dream Machine by M. Mitchell Waldrop
- The Myth of Artificial Intelligence by Erik Larson
- Superintelligence by Nick Bostrom
- Life 3.0 by Max Tegmark
- The Alignment Problem by Brian Christian
- Human Compatible by Stuart Russell
- The Precipice by Toby Ord
- The Bomb by Fred Kaplan
- Command and Control by Eric Schlosser
- The Strategy of Conflict by Thomas Schelling
- The Guns of August by Barbara Tuchman
- The Rise and Fall of the Great Powers by Paul Kennedy
- The Sleepwalkers by Christopher Clark
- The Accidental Superpower by Peter Zeihan
RECOMMENDATIONS- Establish a classified task force to assess and mitigate AGI risks to national security.
- Increase federal R&D funding for AI safety research to $100 billion per year by 2025.
- Overhaul immigration policy to staple a green card to every US STEM graduate degree.
- Harden critical AI infrastructure against cyberattacks and insider threats from foreign adversaries.
- Develop post-quantum encryption standards to protect sensitive data from future AGI capabilities.
- Launch a public education campaign to raise awareness of the transformative potential of AGI.
- Strengthen export controls on semiconductor manufacturing equipment to slow China's AI progress.
- Create an international coalition of democracies to coordinate AGI development and safety standards.
- Increase DoD funding for AI-enabled weapons systems to maintain a strategic advantage over China.
- Establish a national AI research cloud to accelerate US leadership in AI capabilities.
- Pass a constitutional amendment to clarify that AGIs are not entitled to legal personhood.
- Develop AGI oversight committees in Congress with top-secret security clearances and technical advisors.
- Create financial incentives for chip manufacturers to build new fabs in the US.
- Increase funding for STEM education programs to build a domestic pipeline of AI talent.
- Launch a Manhattan Project for clean energy to power AGI development without carbon emissions.
- Establish a national center for AI incident response to coordinate actions during an emergency.
- Develop international treaties to prohibit the use of AGI for offensive military purposes.
- Increase funding for the NSA and CIA to counter foreign espionage targeting US AI secrets.
- Establish a national AI ethics board to provide guidance on responsible AGI development.
- Launch a government-backed investment fund to support promising US AI startups.
ONE-SENTENCE TAKEAWAYThe US must launch a government-led crash program to develop safe and secure AGI before China does.
—
You can create your own summaries like these using Fabric’s extract_wisdom pattern found here.
To learn more about Fabric, here’s a video by NetworkChuck that describes how to install it and integrate it into your workflows.
Powered by beehiiv
June 4, 2024
UL NO. 435: Making New Things is Post-AI Safety
👉 Continue reading online to avoid the email cutoff issue 👈
SECURITY | AI | MEANING :: Unsupervised Learning is my continuous stream of original ideas, analysis, tooling, and mental models designed to help humans thrive in an AI-powered world.
TOCNOTESHey there,
Lots of good stuff this week:
My definition of prompt engineering JUMP
My analysis of AI’s impact on the job market JUMP
I’m working on a number of new talks. I didn’t think I’d be doing this. I planned on having like 2-3 for the whole year, and just doing them different places. But I like this problem. It’s better than the opposite of having no new ideas.
I published a new official_pattern_template in Fabric. It’s basically all my latest format and instruction tricks in one place, and it’ll be the new model I apply to all my prompts. And I’ll keep it updated as the state-of-the-art advances and empirical testing becomes available. MORE
I did my buddy Jason Haddix’s RED BLUE PURPLE AI class last week, and it was phenomenal. Jason is the best teacher I know and going to his classes are a unique combination of learning, hanging with friends, and bettering yourself—all in one. Absolutely loved both days of content. When it becomes available again, you should sign up.
Ok, let’s get to it…
MY WORKCheck out this new sponsored conversation I had with Abhishek Agarwal, co-founder and CEO of Material Security. We talk about:
why product managers are such good co-founders
the need for customized security measures for different organizations
the role of AI in detecting email threats
the importance of single-tenant environments for sensitive customers
and the potential risk of default settings in productivity suites like Google Workspace
Abhishek seriously built this service the same way I would have, and I absolutely loved the conversation.
—
🎙️ My new piece on getting the perfect sound from your microphone, including a bunch of audio tips and analysis of the sounds of top podcasters.
Analysis of Mics and Mic Sounds Used by Podcasters
The difference between different mics, their sounds, post-production, desired sounds, and other podcast-related microphone information
danielmiessler.com/p/podcast-audio
—
This one is political, but centrist. I think you’ll like it if you’re not extreme on either end, but if you don’t like political content—skip it. 😀
The Left's Brexit
The right hobbled the UK with Brexit, and now the left in US has done the same with Stormy Daniels
danielmiessler.com/p/left-brexit
SECURITYCheck Point Vuln
Check Point swiftly released hotfixes for a VPN zero-day, CVE-2024-24919, which is being exploited to infiltrate networks. | HIGH | RESPONSE: Hotfixes released for affected versions. | MORE
Massive Botnet Bust: $5.9 Billion Stolen
The US and international agencies dismantled a colossal botnet responsible for stealing $5.9 billion. This botnet, known as 911 S5, infiltrated over 19 million IP addresses across nearly 200 countries. MORE
AI Against Jamming: Ukraine's Drone Solution
Ukraine's special forces developed Eagle Eyes, a software that lets drones navigate using AI and machine vision, bypassing Russian jamming. Eagle Eyes uses AI to compare live video with pre-collected maps, enabling drones to operate independently of GPS or operator signals. MORE
💡This is the exact tech from Daniel Suarez’ Kill Decision. Basically, offline drones that didn’t need a connection to any network or signal to function.
Sponsor
Enhance Enterprise Security: Trust Every Device with 1Password!
When you go through airport security, there's one line where the TSA agent checks your ID, and another line where a machine scans your bag. The same thing happens in enterprise security, but instead of passengers and luggage, it's end users and their devices.
These days, most companies are pretty good at the first part of the equation, where they check user identity. But user devices can roll right through authentication without getting inspected at all. In fact, 47% of companies allow unmanaged, untrusted devices to access their data. That means an employee can log in from a laptop that has its firewall turned off and hasn't been updated in six months. Or worse, that laptop might belong to a bad actor using employee credentials.
1Password finally solves the device trust problem. 1Password ensures that no device can log into your Okta-protected apps unless it passes your security checks. Plus, you can use 1Password on devices without MDM, like your Linux fleet, contractor devices, and every BYOD phone and laptop in your company.
Visit 1Password.com/unsupervisedlearning to watch a demo and see how it works.
1Password.com/unsupervisedlearning
Watch a DemoSmart Home Tech in Warfare
Home Assistant is being repurposed to alert people of incoming missile and drone attacks in Ukraine. MORE
Data Deletion Services
Incogni is another service like DeleteMe that removes your info from over 170 data brokers. MORE
Sponsor
Build Trust and Accelerate Growth with Vanta
Whether you’re starting or scaling your security program, Vanta helps you automate compliance for SOC 2, ISO 27001, and more.
Streamline security reviews by automating questionnaires and demonstrating your security posture with a customer-facing Trust Center.
Over 7,000 companies like Atlassian, Flo Health, and Quora use Vanta to manage risk and prove security in real time.
Get $1,000 off VantaTicketmaster Breach
Ticketmaster got hit by a group called ShinyHunters, and it looks like around 560 million people’s data has been stolen. MORE
A Few Spread the Most Fake News
In 2020, a tiny group of 'supersharers', mainly older Republican women, spread 80% of the misinformation online. The studies reveal that small, persistent groups can have a significant impact on propaganda and misinformation, and in this case, especially on vaccine hesitancy. MORE
NSA's Easy Hack Protection Tip
The NSA released a guide to keeping your phone from being hacked. One of its top recommendations was simply to reboot your phone once a week. MORE | THE REPORT PDF
👉 Continue reading online to avoid the email cutoff issue 👈
TECHNOLOGYSpeed King: Groq's Llama 3 8B
Groq's Llama 3 8B model exceeded 1,200 tokens per second. I’m telling you, this thing is FAST. It seriously feels fake. It’s like the result is pre-run and stored, but it’s actually running in real time. MORE | GO CHECK IT OUT
💡So here’s the question: why doesn’t Google or OpenAI or Microsoft just offer this guy $1 billion in cash for this technology? Or has that happened already? This is one of the most insane stories in AI right now.
One guy basically has magic beans (proprietary chip tech) that lets him run AI multiple times faster than anyone else in the world.
OpenAI's Apple Deal: A Game Changer?
Satya Nadella is reportedly freaking out over a potential OpenAI and Apple collaboration. Cool story but I just want Apple to fix Siri. If that doesn’t happen, nothing happened. MORE
Outperforming Standard Attention
Researchers have developed new attention mechanisms that actually outperform the standard multi-head attention. MORE
Recall: A Privacy Disaster?
“Stealing everything you've ever typed or viewed on your Windows PC is now a reality, thanks to a feature in Copilot+ Recall that's a privacy nightmare.”
AI wrote that, and while I think it’s true, I disagree with the take. As I talked about last week, this tech is pure magic. In 10 years, almost everyone will think computers without this functionality are worthless. MORE
💡Actually in 10 years most people will be talking and gesturing to their computers. The “computer” will mostly be monitors and/or HUD displays, and your personal AI will be the one doing the work.
So the idea of your AI forgetting work that you did, or who said what when, will be a completely asinine idea. OF COURSE your AI remembers everything! How else would it be useful?
LLMs Aren’t Just Internet Simulators Anymore
LLMs are increasingly being trained on custom, non-internet data, enabling them to exceed the limitations of "internet simulation”. MORE
Nvidia vs. Apple
Nvidia is getting ready to overtake Apple in market cap. Bonkers. MORE
💡Here’s a way to make sense of this:
Imagine that there’s a giant filter on how many people are capable of:
Creating a hit movie
Writing a book
Starting a business
Sharing their art with billions of people
In other words, imagine the current number of major builders/creators is something like .0000000000017% of our 8 billion people. (I made that number up, just work with me.)
The reason NVIDIA is rising so fast is because it’s one of the primary pieces of tech that will remove 3-5 zeroes from that number.
We’re talking about multiplying humanity’s creative output by THOUSANDS in the next decade.
That requires chips.
Thank you for coming to my TED talk about NVIDIA.
Google's Playbook for Keeping Things Simple
Google's SRE Handbook champions simplicity as a core principle for reliability. MORE
💡They need one for product management.
Kaparthy’s GPT-2
🔧 Reproducing GPT-2 (124M) in llm.c in 90 minutes for $20 | by Andrej Kaparthy | MORE
Tinygrad Updates
🔧 Tinygrad's latest update slashes Python time and ditches external dependencies. | by tinygrad | MORE
Terminal Animations That Will Blow Your Mind
🔧TTE: Terminal Text Effects brings a splash of visual flair to your terminal with animations like beams, binary paths, and even a black hole effect. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
HUMANSINR: The Spot-On Predictions Bureau
There's this little-known intelligence bureau that nailed the outcomes in Vietnam, Iraq, and Ukraine. MORE
💡I am going to build some kind of programmatic / AI way to get this group’s predictions into my field of vision.
Sleep's Economic Divide
This study says sleep inequality in the US is deeply tied to economic stress and health disparities, not just bad "sleep hygiene." MORE
Exercise Rewires Brain to Forget Trauma
Mice with PTSD-like behaviors showed significant improvement after having access to a running wheel, hinting at exercise-induced neurogenesis as a potential therapy. MORE
Taste the Difference: How Ozempic and Wegovy Could Change Eating Habits
Ozempic and Wegovy might be fine-tuning your taste buds to help you lose weight by making sweets taste sweeter. MORE
Carmack Questions Work's Worth
John Carmack dives into "Bullshit Jobs," sparking a lively debate on the value of work. Carmack's review sheds light on the often unspoken reality of perceived value versus actual productivity in modern employment. MORE
The Social Lives of the Teens Who Don't Have Phones
“Imagine navigating high school's social maze without a smartphone, missing out on chats and memes but gaining a unique perspective on life and friendships.” MORE
💡I don’t have to imagine; I grew up in the 80’s. It was glorious.
Could Eye Exercises Be the Key to Fighting Myopia?
A comprehensive review finds eye exercises might actually help in preventing and controlling myopia. MORE
Lung Cancer Breakthrough: Lorlatinib's Unprecedented Success
A new lung cancer drug, lorlatinib, has shown unprecedented success, keeping 60% of advanced-stage patients progression-free for five years. MORE
Marc Andreessen Wants You to Stay in School
Marc Andreessen's advice to a Stanford student to "stay in school" is surprisingly grounded. The real kicker is his reasoning: if you're the type to drop out, you probably wouldn't listen to advice to stay anyway. MORE
I Love My Wife. My Wife Is Dead.
Richard Feynman's undelivered letter to his late wife reveals a heart-wrenching blend of love and grief, even years after her death. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
IDEAS & ANALYSISHere’s my new favorite way to explain prompt engineering. In short, don’t think of it like an AI thing, or a tech thing. Think of it this way instead.
—
My X thread on why it’s so hard to find tech jobs right now, and how I think AI is affecting the situation:
The problem for the job market isn't that AI is happening. The problem is that it’s happening at the exact same moment that most companies are figuring out that 80% of their employees are worthless.
We need to stop expecting things to go back to the way they were.
The new reality is companies mostly hiring super ambitious, exceptional, proven people who are gods with AI.
This means most formal education becomes a waste of time and money because a degree doesn’t certify you are any of those things.
In this model, only elite schools will matter because the filtering for being exceptional will have happened just by being accepted into the school.
It’s like two separate things:
1. You’re exceptional enough to be accepted here.
2. You finished some classes.
#2 you can get anywhere. #1 is the thing that employers actually care about.
So the result will be companies hiring the top 10-20% of people in competence. This will be filtered by:
1. Elite school attendance (if young)
2. Proof of competence via something you’ve put into the world, like on your website, YouTube, as a tool, or as a company you built.
So, in hyperbolic form:
If you’re not 19 and at Harvard—or have your own projects or companies you’ve built and talked about online—you are not going to be interesting to employers.
You’ll be part of the 90% fighting for scraps.
RECOMMENDATION OF THE WEEK
The problem for the job market isn't that Al is happening.
The problem is that Al is happening at the exact same moment that most companies are figuring out that 80% of their employees are worthless.
We need to stop expecting things to go back to the way they were.
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
Jun 2, 2024
For everyone you know who is going to be thinking about their career in the future—which I guess means everyone who’s not independently wealthy—try to get them thinking in the following way:
AI is going to take out most executors of knowledge work.
The people who will survive are the people with ideas who are actively. building/making that thing.
So the safest thing to be is a CREATOR or a BUILDER.
That means founder, entrepreneur, programmer with ideas, productive and ambitious artist, etc.
The trick is you have to be able to 1) have the ideas, 2) make it yourself or get the talent/AI together to have it made, and then 3) market the hell out of it.
In short, the winners in this new game will be the people making new things and bringing them into the world.
APHORISM OF THE WEEKThank you for reading.
UL is a personal and strange combination of security, AI, tech, and lots of content about human meaning and flourishing. And because it’s so diverse, it’s harder for it to go as viral as something more niche.
So—if you know someone weird like us—please share the newsletter with them. 🫶
Share UL with someone like us…Happy to be sharing the planet with you,
Powered by beehiiv
June 3, 2024
Analysis of Mics and Mic Sounds Used by Podcasters
I think I’ve finally figured out what’s up with podcast audio. At least in terms of:
What makes a good microphone
What makes a good sound signature in a podcast
Audio production
Etc.
So what I’m going to do is just blast you with what I at least think I know at this point.
My audio journeyI’ve been studying this stuff in-depth since around 2019. And during the pandemic I went hard on it. Like full audiophile.
I learned the science around what audiophiles were on about. I learned about what makes good sound in music. I built my house in 2020 and put in a pretty sick sound system.
But a lot of my focus was on finding the perfect sound for a podcast. Like how I wanted to sound in audio, and what it takes to make that happen.
Previous postsMy podcast sound update in 2021 MORE
Chasing the perfect microphone sound MORE
This post basically updates and replaces a lot of what I talked about in those, although much of what I said before still holds.
The basics, in no particular orderFirst I’ll hit you with the science / reality stuff and then I’ll give you some samples.
Dynamic microphonesDynamic mics have less range but are more forgiving of noises.
The Shure SM7b is the most famous dynamic mic right now, and pretty much every major podcaster uses it.
The SM7b produces a pleasing, rounded, deep, but a little bit dead sound. It makes you sound authoritative.
Condenser microphonesCondenser mics are extremely clear but very noisy because they pick up everything.
Condensers capture mouth noises, any room sounds, and pretty much anything VERY clearly and accurately.
They also pull in much more sound range, meaning way up towards 20,000 hz and beyond, which makes them sound less dead and more realistic.
Examples include the Yeti series on the low end (less than $100), and the Neumann U87ai on the high end ($4,500).
Common mistakes with post-productionLots of podcasters—especially men—like a super-deep sound because they think it makes them sound cool and authoritative.
Unfortunately, all that extra bass sounds artificial over time and is actually quite fatiguing to listen to.
It’s also really hard to listen to in a car, where lots of people listen to podcasts, because the low rumbling mixes with the road noise.
Signature soundsThe NPR sound is low bass and high clarity largely because they use Neumann U87ai mics, and they like that clear sound.
The sounds of famous podcastersSam Harris uses an SM7b (or the USB version) and keeps it pretty clean with not too much post-production. He’s had the same signature for years now. It’s basically mid-range, with some decent bass but not too much, extremely clean in terms of mouth noises and background noises, and a bit sterile. It works well for him because it’s like an understated clean sound.
Chris Williamson is a super-popular podcaster now and he also uses SM7b’s, but he mixes his own mic with a lot of post-production to have lots more highs and “ambience” added in. Which basically means character. If you listen to his podcast his voice always has a strong combination of deep but also bright. Really good production, and I’d be surprised if he or his team ever shared his post-production chain. I bet he considers it a trade secret.
Joe Rogan also uses the SM7b and he has a sound profile that’s very neutral and natural. I’m not sure if Jamie enhances the highs or not, but the overall mix is very clean and clear.
Lex Fridman is also on an SM7b and he has a pretty neutral sound with some highs possibly added in.
I believe Scott Galloway is on SM7b’s as well, and he definitely has some added in highs after the fact. His sound is a lot more natural and bright than something like Joe’s.
Tips for improving your audioThe general rule for audio is to do as much as possible at recording time and as little as possible in post-production. That means 1) room treatment, and generally having a quiet room if possible.
Removing echo is a superpower for making audio sound amazing. Definitely do this if you want to sound professional. I have been using DeVerberate for years and love it.
If you’re going to use the SM7b, my recommendation is to use the it with bass rolled off and the highs raised on the microphone itself. This will still have plenty of bass but it’ll make the mic sound much more natural and vibrant vs. sterile and dead.
Do as little post processing as possible.
Treat your room if possible.
Make sure to remove echo, even if you don’t think you have much.
Additional trivia / tricksLoudness is deceptive and an enhancer. Basically, being louder sounds better to most ears, but be careful with how you achieve it.
Be careful with loudness without compression because you can start clipping (cutting out from being too loud).
Be careful with too much compresssion because it can remove the natural sound and make it sound more sterile and dead.
My current setupMy current setup is quite clean and simple. I’ve moved away from my Neumann U87ai because it picks up too much noise. Most importantly my mouth noises, shaking feet, the sound of my clothing rubbing, etc.
SM7b
Fully treated studio / office where I work
Bass rolled off (on the mic hardware)
Highs enhanced (on the mic hardware)
RODECASTER PRO for my mixer
The only mod on my RODECASTER PRO is a noise gate
Post-production when I record in Hindenburg and not OBS:
DeVerberate for echo removal
Nectar 4 for noise gate
Nectar 4 for bass cutoff below 80
Nectar 4 for a slight lift in mid bass
Nectar 4 cutout at around 1500 hz to remove bad mids
Nectar 4 minor high shelf added after around 7000
Here’s what that sounds like
// My latest mic sound
SummaryYou should probably use a dynamic mic unless you’re a trained studio speaker.
You should probably use a dynamic mic if you don’t have a treated room.
Resist the urge to add bass. It only sounds cool for a moment, and then is bad for listeners.
If you have any doubts, go with the SM7b. It’s a workhorse. And if you can’t afford one yet (around $400), get the little brother USB version.
If you do get an SM7b you’ll need an audio interface to put it in, and/or a signal booster.
Focus on getting a clean, natural sound from your voice, not making it sound deep or booming. That is what will cut through the listener’s environment and make it to them the best.
Do as much as you can during recording, and try to modify it as little as possible in post.
Finally, don’t stress it too much. The content matters more anyway.
Powered by beehiiv
May 31, 2024
The Left's Brexit
I think Trump's conviction might be the American Left's version of Brexit.
Let me explain.
First off, I think Trump is the worst president we've ever had by far, and I do NOT want him re-elected. I am anti-current-Left, but I consider myself a Radical Centrist and have never voted for a Republican. So that's where I am politically.
I also think that in normal circumstances it would be ok to prosecute someone for what Trump did. He appears to be guilty of the crimes, and he's now a convicted felon as a result. Fine.
My problem is how and when it happened.
First, a lot worse crimes have been carried out by presidents without prosecution. Like, really bad stuff (See: spending trillions in Iraq to remove someone protecting the world from Iran, the Iran/Contra issue, and many others). So, we absolutely set a precedent here by doing this.
Second, it seems entirely too obvious why it happened. Basically, the radical left is scared Trump will be re-elected, and none of their previous (and legitimate) weapons were effective at taking him out, e.g., impeachment, Jan 6 hearings, etc.
So basically—after everything that's happened—January 6th included—at least half the country still wants to elect Trump. And the Left is VERY angry and scared as a result.
So they did the Al Capone Attack. Basically, hit them for something else they did that was illegal. And that was this Stormy Daniels thing. A last-ditch, desperate effort to take down someone who they don't believe should be President again.
Here's the problem. They don't get to fucking decide that. Half the country has seen all this evidence and they still want him. This shakes me deeply. I don't see how that's possible. But it's true. Go check the polls. It's tied. Half. Half the country still wants him over The Left.
Actually, I eventually did figure this out. Most people voting for him also hate him compared to a theoretical, better conservative, but they love him because he's the best weapon against the Radical Left. In other words, they love him because he's their defense against the Left turning the country into San Francisco.
Anyway. They want him. That's the reality. Half the country. And the Left is panicking.
My problem with this whole fucking thing is that this is self-inflicted by the left. Just like Brexit was for the right in the UK. The issue is the left is NOT LISTENING to how much America despises far-left policies. The country keeps communicating very clearly that they are center-right. That they're more conservative.
And the left ignores it. Over and over. And instead of offering logical, progressive, centrist, UNIFYING policies that appeal to everyone, they repeatedly go Extreme Left and alienate the country.
And they wonder how a clown like Trump is so popular. Like it's a fucking mystery.
Let me spell this out for you, friends on the left. Your extreme policies have made you the #1 enemy of America, which has forced them to choose ANYONE to oppose you—including someone they hate—like Donald Trump.
In short, this is all the left's fault. All of it. Trump being elected the first time. And if Trump gets elected again, that will be their fault as well.
Self-inflicted. Just like Brexit.
So, to my friends on the left, expect more Trump types unless you stop kicking balls in your own goal and START LISTENING.
You have a choice: either wake the fuck up or stop being surprised when bad things happen.
Powered by beehiiv
May 28, 2024
UL NO. 434: Can You Articulate Yourself in 50 Words?
👉 Continue reading online to avoid the email cutoff issue 👈
SECURITY | AI | MEANING :: Unsupervised Learning is a website, newsletter, and podcast about how to survive and thrive as humans in a post-AI world.
TOCNOTESHey there,
I think one of the most important things for surviving AI is being able to articulate what you do. Or—even better—who you are.
I was at a party with a bunch of muckety-mucks recently and someone asked me what I did. I told them I built AI stuff, and it was quite unsatisfactory—both to me and to the person I gave the answer to.
Here’s how I’ve decided to answer from now on:
I’m basically worried about two problems: people having a lack of meaning in their lives, and what will happen to peoples’ sense of meaning when AI takes their jobs.
So what I do is use AI to build products and services that help people and companies create a version of themselves that will thrive after AI is everywhere.
My new spiel
I like this for a few reasons:
It starts with the problems
It highlights that I’m building stuff
It mentions AI, which is what I’m doing
It puts meaning as the most important piece, above AI
As I’ve said before, I think the most screwed people are those who don’t know themselves. People without a purpose that they’re driving towards. And even people who have a purpose but can’t articulate it.
Know yourself. And be able to articulate it.
Oh, and the create_5_sentence_summary Fabric pattern is super helpful here. Check this out:
I’m in love with using AI to give Accordion Explanations to things. Expand, compress. Expand, compress.
Ok, let’s get to it…
MY WORKMy buddy Network Chuck just released a video on Fabric! It’s easily the best video on the tool, including all the ones I’ve made myself. Go check it out! WATCH THE VIDEO
SECURITYAnd Now a Cyber Force
The House Armed Services Committee is preparing to propose a study for a dedicated military Cyber Force, echoing the creation of the Space Force. MORE
China's ORB Networks
Chinese-linked hackers are getting harder to find due to 'ORBs'. These ORB networks, not directly run by the government but by contractors in China, are a maze of compromised devices and virtual servers used for spying. MORE
Fix Problems Instead of Doing Risk Measurement Theater
Andy Ellis suggests fixing tangible security problems might be more productive than obsessing over measuring risk. Love this take. MORE
💡So I guess there would still be cyber within each group, but this would be a dedicated branch. It would be pretty difficult to have all cyber people belong to this group and have them assigned to different branches. I bet it’ll just be a dedicated one separate from cyber within each branch. Cool either way, though.
Sponsor
Secure Email Beyond Just Phishing
There’s more than one way-in to exploit email as an attack vector. Plus even more to target once inside the mailbox.
Material takes a holistic approach to email security that covers the full threat landscape – stopping new flavors of phishing and pretexting attacks in their tracks, while also protecting accounts and data from exploit or exposure.
Visit material.security to learn more about their multi-layered detection and response toolkit for email.
Secure Email From Every AngleCritical Bug Exposes Major Clouds
A critical vulnerability named "Linguistic Lumberjack" in Fluent Bit was found in a utility with over 3 billion downloads. It creates risks of denial of service, data leaks, and remote code execution on systems including AWS, Microsoft, and Google Cloud. MORE
American Criminal Records Exposed
EquationCorp and USDoD (threat actors), just dumped a 70 million row database of American criminal records. MORE
Space Warfare Escalates
The U.S. accuses Russia of launching a space weapon capable of attacking satellites, which Russia dismisses as "fake news." Despite the denial, the Pentagon asserts the weapon is in the same orbit as a U.S. government satellite, posing a direct threat. MORE
Sponsor
Build Trust and Accelerate Growth with Vanta
Whether you’re starting or scaling your security program, Vanta helps you automate compliance for SOC 2, ISO 27001, and more.
Streamline security reviews by automating questionnaires and demonstrating your security posture with a customer-facing Trust Center.
Over 7,000 companies like Atlassian, Flo Health, and Quora use Vanta to manage risk and prove security in real time.
Get $1,000 off VantaAI Breaks Into Tech Giant
IBM's X-Force used their new AI hacking platform to breach a major tech manufacturer's network in just 8 hours. This will only get better. I’m doing a video on this soon. MORE
Tesla's High-Tech Keys? Still Hackable.
Despite Tesla's new ultra-wideband keyless tech, a simple radio hack can still steal these cars. MORE
NTSB Preliminary Report on Baltimore Bridge Collapse Released
The NTSB's initial findings on the Baltimore bridge collapse hint at unexpected structural weaknesses. The report suggests that overlooked maintenance issues could have contributed to the sudden failure. MORE
Core WiFi Bug Discovered
Researchers have discovered a new WiFi vulnerability affecting all clients and operating systems, due to the IEEE 802.11 standard's lack of SSID authentication during beaconing. MORE
US Scheme Funds North Korean Nukes
A U.S. woman and four others were charged in a scheme using fake IT jobs to funnel millions to North Korea's nuclear program. The U.S. Justice Department's indictment reveals a sophisticated operation that compromised over 60 U.S. identities and impacted more than 300 companies, generating at least $6.8 million for overseas IT workers. MORE
NYPD's Drone Response Team
NYPD's rolling out drones to respond to 911 calls in five NYC precincts. I love this more than I should. It’s super efficient and sci-fi at the same time. With a background of dystopia. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
TECHNOLOGYThe EU Passed Its AI Act
The AI Act has been passed and my general opinion is that it’s good content but I worry it will be used to keep the EU sidelined in the race to AI. Here’s a good summary and browser of the actual act. MORE
Main Points:
1. Prohibits certain unacceptable AI practices that are manipulative, exploitative or cause harm.
2. Classifies certain high-risk AI systems that pose risks to health, safety or fundamental rights.
3. Establishes requirements for high-risk AI systems related to risk management, data governance, technical documentation, transparency, human oversight, accuracy and cybersecurity.
4. Sets obligations for providers and deployers of high-risk AI systems.
5. Requires conformity assessments, registration, and post-market monitoring for high-risk AI systems.
6. Mandates transparency obligations for certain AI systems like emotion recognition and biometric categorization.
7. Provides measures to support innovation, especially for SMEs and startups, like regulatory sandboxes.
8. Establishes a governance system at EU and national levels, including a European Artificial Intelligence Board.
9. Allows for fines up to 7% of worldwide turnover for non-compliance with the rules.
10. Aims to facilitate development of codes of conduct for voluntary application of requirements to non-high-risk AI. MORE
When Your AI Sounds Too Much Like Scarlett Johansson
OpenAI pulled the voice that sounds like Scarlett Johanssen, and it’s turned into a giant mess that’s challenging OpenAI’s morality. Specifically, Sam’s. MORE
💡I have opinions on this. Basically they knew they wanted to make it sound like the AI from Her. That’s why Sam tweeted out, “Her.”
So it seems like they basically tried to make that happen two ways: 1) with a separate voice actress who sounds similar, and then by reaching out to Scarlett herself. Scarlett said no. So somebody there tried to pivot and say it wasn’t supposed to sound like the movie.
I don’t think this is nearly as malicious as people are making it out to be. Remember that things get crazy inside companies when there are multiple parties both taking actions and talking to the media. I think it’s more like good intentions and bad execution. But I do think it’s a mess. And a self-inflicted one.
The irony is that this is doing more damage to the appearance of Sam’s morality than the ouster by the board. And again—self-inflicted on OpenAI’s part.
Apple Locks Down 2nm Chips
Apple apparently snagged all of TSMC's 2nm chip production capacity in a fairly private meeting. MORE
Nvidia's AI-Driven Revenue Surge
Nvidia's revenue skyrocketed by 262% due to the AI frenzy, hitting $26.0 billion and surpassing expectations. MORE
US Strikes Lithium Gold
The US just found a massive, untapped lithium source. MORE
Windows Reimagined: AI at the Core
Microsoft is reimagining Windows with a deep dive into AI, introducing Copilot+ PCs that blend cutting-edge AI features directly into the operating system. MORE
💡People are freaking out and saying nobody wants this. Yes. Yes they do. Once everyone has it, it’ll be crazy to imagine a time when they didn’t have it.
This is approaching the ideal form factor of a computer. Like Her, actually. Anything you do on your computer, or in your life, is available to your assistant. And you can ask it anything or have it collaborate with you on work. WATCH A WHOLE VIDEO ON THIS
Anything less is less than complete. The faster we accept this as the future of computing the faster we can move on.
Will there be security and privacy issues? Yes. SO MANY. And they’ll be horrible because it’ll be your whole soul getting compromised. But it’ll be so useful that people will happily enable the functionality anyway. WATCH A WHOLE VIDEO ON THIS
You might not believe me, and you might abstain yourself. But most will jump in head-first as soon as they get a taste of the capabilities.
Mapping AI's Inner Universe: A Peek Inside Claude Sonnet
Anthropic released a brilliant paper that shows how one of its models think. It’s a major step in AI explainability, and if you care about this field at all you should check out this paper. MORE
New Chinese Chatbot: Ask President Xi
China's new chatbot lets citizens ask President Xi how to be better socialists, trained on his philosophies and major internet databases. MORE
AI: The New Astronomer's Assistant
Astronomers are preparing to use AI to tackle 300 petabytes of data annually. Cecilia Garraffo's AstroAI initiative is pioneering the fusion of AI and astronomy to explore deep cosmic questions, already planning dozens of projects with a 50-member interdisciplinary team. MORE
Windows now has AI-powered copy and paste
Windows 11's PowerToys now lets you AI-transform clipboard content into different formats or languages with a simple shortcut. This Advanced Paste feature can do everything from converting code between languages to changing text style, but it hinges on having OpenAI API credits. MORE
Don't Microservice, Do Module
Microservices might be all the rage, but this deep dive suggests we're better off building with modules instead. MORE
No More Public Likes
X is ditching public "likes" to stop people from holding back on liking edgy content due to fear of backlash. MORE
The Pragmatic Programmer
Focusing on fundamentals over frameworks could be the secret sauce for long-lasting tech skills, according to the Lindy effect. MORE
🔧 This guide shows you how to use Ghidra for crafting a GDB script that traces function calls, making vulnerability hunting a bit less of a headache. | by Craig Young | MORE
🌐D3 in Depth guides you through crafting interactive data visualizations on the web using D3.js, Chart.js, Leaflet, and React. This resource not only covers D3.js but also dives into the essentials of HTML, SVG, CSS, and JavaScript for creating compelling data stories and visualizations. By Peter Cook | MORE
🛠️Experts.js simplifies creating multi-agent AI systems with OpenAI's Assistants API, enabling complex tasks through a panel of specialized tools. By Metaskills | MORE
🔧 The "User Persona Generator" Prompt simulates user interviews to create detailed user personas, saving you from conducting dozens of real interviews. | by The Prompt Warrior | MORE
👉 Continue reading online to avoid the email cutoff issue 👈
HUMANSJob Market Paradox
Job seekers are struggling to find work, and employers can't seem to fill positions. This paradox highlights a significant mismatch in the job market, possibly due to skills gaps or unrealistic expectations on both sides. MORE
💡It’s not a mystery: the problem is that hiring managers need people ready to do the job, and they don’t want to train anyone. This is why degrees are mattering less, and why someone with an active tech blog or YouTube channel—who is actually building stuff and doing the thing—can get hired with no credentials over someone with a degree in computer science.
The ability to actually do the thing is all that matters now.
Expect to see AI services soon that look at a person’s social media, blog, YouTube, and then do an interactive interview with them, and they give a score of how likely they are to thrive in a given job.
This type of algorithm will cut through credentials, fluff, and all sorts of other BS and get at the real thing.
The value of traditional education will drastically fall soon as a result. Because degrees won’t significantly raise those scores for most people. And for the people who are likely to have high scores, the degree won’t raise it that much.
Basically it’ll filter for drive, curiosity, obsession, and proof-of-work. Like actual builders and coders. And people who raise insightful questions and are curious.
This means that not only will most candidates fail the test, but over 90% of college graduates will as well. So college will largely become this very long and expensive waste of time.
It’ll spawn a new type of education dedicated to making more curious and capable people.
Tough Times for New Grads
Recent grads are hitting a wall with entry-level hiring projected to drop by 5.8% in 2024, making the job hunt tougher than it's been in years. Software engineer positions on Indeed have plummeted 30% from pre-pandemic levels, signaling a significant shift in the tech job landscape. MORE
💡See above. And of course this isn’t the only factor; I’m sure there are other causes as well.
Is College Overrated?
Nearly 30% of Americans are questioning the value of college, according to a recent Pew poll. MORE
💡See above.
Job Hopping: Not Just Flaky, But Strategic
High performers often job hop not because they're flaky, but because they're in search of a culture that matches their drive. Sticking it out in a toxic work environment can dull a high performer's edge, making them lose their superpowers over time. MORE
💡For truly top performers, it’s not that bad to see them jump every 3-4 years. Hiring managers know they get bored quickly and that they probably solved a bunch of problems and wanted a new challenge.
Sunlight: A Double-Edged Sword?
So now they’re saying avoiding the sun might actually be a risk factor for major causes of death, sending mixed messages on sun exposure advice. MORE
💡I definitely believe this. I think there’s value in this whole “get sun, touch grass” thing. Grounding. That kind of thing. Like a more, um, grounded version of the “get in touch with nature vibe”.
I can tell you it’s true for me anecdotally, and that it seems to be true anecdatally. Would love to see some more real science on it though.
Daily Marijuana Use Outpaces Daily Drinking in the US, a New Study Says
For the first time, daily and near-daily marijuana use in the U.S. has surpassed high-frequency drinking. MORE
Psychedelics as Painkillers?
Psychedelics are being eyed as a potential boon for treating severe chronic pain, including cluster headaches. Research is exploring LSD and psilocybin, not just for their mind-altering effects, but for their ability to significantly reduce debilitating pain. MORE
Amazon's Surprising Stat
One-third of Amazon warehouse workers rely on food stamps or Medicaid, which is quite the eyebrow-raiser. MORE
One File to Rule Them All
Imagine keeping all your knowledge nuggets in one massive, sprawling text file—simple, yet surprisingly effective. This approach turns the digital clutter of countless notes and documents into a singular, searchable, and highly efficient knowledge base. MORE
Breaking Up the Big Show
The US Justice Department is gearing up to dismantle the Live Nation-Ticketmaster giant. So overdue. MORE
California Solar Over 25%
California's solar power has hit a new milestone, now powering over 13.9 million homes and making up more than a quarter of the state's electricity. MORE
Sword Fighting is Not What You Think…
An argument from one expert claiming everything we think we know about historical sword fighting is pretty much wrong. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
IDEAS & ANALYSISMy Favorite Paul Graham Essay
This is my favorite part of my favorite essay from Paul Graham:
Great work usually entails spending what would seem to most people an unreasonable amount of time on a problem. You can't think of this time as a cost, or it will seem too high. You have to find the work sufficiently engaging as it's happening.
There may be some jobs where you have to work diligently for years at things you hate before you get to the good part, but this is not how great work happens. Great work happens by focusing consistently on something you're genuinely interested in. When you pause to take stock, you're surprised how far you've come.
Paul Graham
This rhymes with a realization I’ve been having lately that the best leaders and founders have something in common: obsession.
It’s not just that they’re disciplined. They might be. And probably are. But they might just be obsessed, which tends to look a lot like discipline because they both produce consistent effort in a direction.
I find myself increasingly looking for people who are obsessed with things. Life. An idea. Whatever. They can’t stop thinking about it, and the million different ideas they have, tend to revolve around that same concept.
—
Most people think life when they were around 10 was the best. Thanks to Joel Parish for the link.
RECOMMENDATION OF THE WEEKBecome able to articulate who you are and what you’re about in 25-50 words.
APHORISM OF THE WEEKThank you for reading.
UL is a personal and strange combination of security, AI, tech, and lots of content about human meaning and flourishing. And because it’s so diverse, it’s harder for it to go as viral as something more niche.
So—if you know someone weird like us—please share the newsletter with them. 🫶
Share UL with someone like us…Happy to be sharing the planet with you,
Powered by beehiiv
May 20, 2024
UL NO. 433: China's Flawed Strategy
👉 Continue reading online to avoid the email cutoff issue 👈
SECURITY | AI | MEANING :: Unsupervised Learning is a website, newsletter, and podcast about how to survive and thrive as humans in a post-AI world.
TOCNOTESHey there,
Hope you’re doing well!
We added a new Fabric Pattern called explain_terms, which takes an input, finds the difficult terms, and gives a definition, an analogy, and why it matters. MORE
Had a great weekend at my annual EDC pilgrimage of music and friendship at EDC in Vegas. So tired. So refreshed.
If you don’t have one or two friend trips planned per year, I recommend you get them going. Even if it’s local camping or something. Just get your friends together away from devices to enjoy each others’ company, music, nature, or whatever.
I’m thinking of writing a short book about what people will need to be anti-fragile in a world full of AI. My working title is: Human 3.0 — The skills and mental frames required to thrive in a post-AI world. Let me know your thoughts on the title, and if you’d be interested in reading a book like that.
Ok, let’s get to it…
MY WORKSECURITY
I'm pleased with how well this AI Attack Surface Map has held up since May 2023.
I still think this is the clearest way to think about real-world risks to AI applications.
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
May 14, 2024
FBI seizes hacking forum BreachForums — again
The FBI, along with international partners, seized BreachForums again—this time also hitting its Telegram channels. They were hit last year too, but bounced back under new management. MORE
Chinese Brain Drain
Microsoft is moving nearly 10% of its China-based tech talent to relocate, signaling a potential tech exodus amidst U.S.-China tensions. This move could lead to a significant brain drain from China, especially since these employees are in the critical fields of AI and Cloud. MORE
💡I am very happy about this.
This is a bigger point than just that story, but it seems like the US is set up to win in so many ways right now. Which is weird, because the US is a mess too.
Europe’s economy is suffering from missing the innovation boat because they’re obsessed with regulation, and China is making their country so hostile that the West is pulling away from it, while the smartest Chinese find ways to get away.
This leaves the imperfect US (and Canada) as the only ones leaning into pure innovation and attracting talent. Nowhere is this more clear than with the Bay Area AI boom.
China is at a major disadvantage with AI because government control over its population depends on a government-controlled narrative that falls apart under scrutiny and analysis.
I feel like AI somehow encourages people to question things, and push what’s possible. To grow, and build, and create. And if the smartest Chinese do that in China they’ll run into obstacle after obstacle. And eventually they’ll just leave, which is great for the West and horrible for China.
So between China hacking everyone, threatening everyone, not having enough kids, and controlling narratives and innovation, they’re potentially building their own implosion.
The world will pull away from them, and their best minds will flee.
Autonomous F-16 AI Pilots
Air Force Secretary Frank Kendall recently flew in an AI-controlled F-16, finding its capabilities nearly on par with experienced human pilots. MORE
Sponsor
Mythbusting Quiz: Is ITOps Automation Safe?
‘IT automation’ has reached buzzword status in the IT community, propelled by its promise to streamline operations, reduce manual errors, and enhance efficiency across diverse IT environments. Automation is defined as the use of software to create repeatable instructions and processes to replace or reduce human interaction with IT systems. However, automation in IT is not just about saving you time; it's also about performing tasks more consistently and rapidly, enabling you to focus on more strategic work.
The hype around IT automation has given rise to numerous myths. We’ve compiled a list of some of the most common myths we’ve seen. Do you think you can tell what’s true and what’s false?
discover.automox.com/ul/2024-it-automation-quiz
Take the Quiz NowSponsor
Secure Your Assets with ProjectDiscovery
Discover ProjectDiscovery, the ultimate platform for vulnerability detection and management. Our cloud-based solutions empower your security team with automated, continuous scanning and seamless integrations. From AI-generated templates to CI/CD implementations, our tools ensure your tech stack is protected against the latest threats. Join thousands of developers who trust ProjectDiscovery for robust, scalable, and open-source security. Stay ahead of vulnerabilities and secure your critical assets today.
Learn More Today👉 Continue reading online to avoid the email cutoff issue 👈
TECHNOLOGYGoogle I/O 2024: everything announced
Everything announced at Google I/O. MORE
It’s just impossible for me to get excited because they’re so bad at product management. They’re like an idea lab instead of a company.
So I’m like, “Wow, that’s cool. Can’t wait for OpenAI or Apple to productize that!”
A lot of the AI stuff Google showed was cool. But I find myself unable to care.
- Their products are far harder to use
- They have a “throw it at the wall” feel
- If I like it they’ll probably cancel it
They’ve simply lost my trust.
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
May 14, 2024
Avoiding the Mediocre Success Trap
Interesting idea about how a home run or striking out is far better than the limbo of a mediocre success. It’s about getting a strong signal, which partial wins don’t really give you. MORE
📃Ilya Sutskever suggests mastering these 30 papers to grasp 90% of crucial machine learning knowledge. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
HUMANSAncient Stars in Our Galactic Backyard
MIT researchers stumbled upon three ancient stars, some of the universe's oldest, casually orbiting our Milky Way. MORE
Whale's Secret Code
Scientists just found that whales have their own 'alphabet' in their songs. My buddy Marc has been working on deciphering this stuff for years. Super interesting. MORE
Swiss vs. American Parenting: A US Mom's Perspective
A US mom shares how Swiss parenting's emphasis on independence and freedom was a culture shock compared to American ways. In Switzerland, kids as young as four are encouraged to walk to school alone. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
IDEAS & ANALYSISSmall Therapy?
I think the magic amount of therapy (for most people—not all) might be some relatively small amount that untangles your knots, but doesn’t point your lens permanently on yourself and your troubles. I feel like too much therapy has the approach of bad Chiropractors—where the goal is more visits, not on getting to where you don’t need the visits.
RECOMMENDATION OF THE WEEK
Interesting idea I just had about therapy:
The people not getting any might need some, and the people getting a lot might be getting way too much.
I think there’s probably some magical, moderate amount that sits squarely between Untangled Knots and Rumination.
🤔
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
May 16, 2024
Make sure you have 1-2 friend trips planned per year. Even if it’s not everyone, and even if it’s not some extravagant thing.
Like 2-3 days. Camping. HIking. Whatever. Even just a role-playing weekend away from computers with friends.
Maybe have a rule of no tech during the time, with the focus being on friendship and the activities you’re doing together.
Super replenishing.
APHORISM OF THE WEEKThank you for reading.
UL is a personal and strange combination of security, AI, tech, and lots of content about human meaning and flourishing. And because it’s so diverse, it’s harder for it to go as viral as something more niche.
So—if you know someone weird like us—please share the newsletter with them. 🫶
Share UL with someone like us…Happy to be sharing the planet with you,
Powered by beehiiv
May 14, 2024
UL NO. 432: Can You Summarize Your Work in a Sentence?
👉 Continue reading online to avoid the email cutoff issue 👈
Unsupervised Learning is a Security/AI newsletter about how to transition from Human 2.0 to Human 3.0 in order to thrive in the post-AI world.
TOCNOTESHey there,
Lots of stuff!
New Fabric Pattern: get_wow_per_minute
This brand-new Fabric Pattern allows you to figure out the value density of any piece of content, rated from 0 to 10.
OpenAI’s Event
OpenAI released their new model, GPT-4o (the “o” stands for omni)
The big news is that it’s just about as smart as 4, but it’s 4x faster and 2x cheaper.
It’s also capable of having real-time conversations in a very realistic way, allowing you to use it for real-time translation and other stuff.
I did a predictions post that anticipated more agent stuff besides just the DA component, but alas we’ll have to wait for that…
It’ll soon have vision as well, so you’ll be able to have it watching your screen, and you can just ask it questions, and it can help you.
They released a desktop app as well, which is where that functionality will eventually live. Only a few people have it so far but I’m evidently one of the lucky ones…but it doesn’t have the screen monitoring piece yet.
💡They’re basically working to create the Digital Assistants I talked about in my Predictable Path video, which is the most obvious but awesome move ever. Tons of people are doing it, but it’s great to see OpenAI jump ahead in this space.
RSA
So, RSA was really good. Like really, really good. Caught up with so many friends and had so many nourishing conversations. Lots of them were about trying to convince friends to get out of their jobs where they’re unhappy, and to get into something AI.
Did a couple of talks and a couple of panels. Lots of fun there.
Probably should have brought something to sell, and had like a sales pitch, but I just hate that vibe. I think I’ll just buy marketing / ads and/or pay sales people for that so I don’t have to do it.
I did show some people Threshold though, which was a total hit.
The coolest thing I saw at RSA was the energy. We’re definitely back. And when I say we, I mean optimism and energy around security/entrepreneurship. Not exactly sure about the mix there, but it definitely felt lively to be around the conference in a way that hasn’t been true in like 5-6 years.
Speaking of Threshold, holy crap! I am LOVING this thing. It’s now my #1 way of finding my highest quality content. Plus we send out an email every day with your feed in case you didn’t get a chance to check it. Here’s my latest one:
My personal Threshold feed
I am not joking when I tell you that every single one of these were hits for me. Every. One. I can’t believe we’ve actually built the content discovery tool that I’ve wanted my whole life. And we’re just getting started with the features. The stuff coming in a few weeks will be insanity.
You should get it. Oh, and UL Members get half-off the first year with a code. It’s in UL chat once you sign up.
Ok, let’s get to it…
MY WORKCheck out this new sponsored conversation I had with Corey Ranslem, CEO of Dryad—and the resident expert on Maritime Attacks—and Ismael Valenzuela, VP of Threat Intelligence and Research at Blackberry.
We talked about all things Maritime Security, and I learned a whole lot from the conversation.
—
Not my work, but my Dad just went on a live studio podcast in San Francisco. He talked about how he approaches music, and performed multiple songs. Go check it out! THE FULL SHOW | A CLIP OF HIM PERFORMING HIS SONG: CHILDREN OF THE NIGHT
SECURITYDell got big-hacked (49 million accounts) by someone scraping an API. One message I’ve not heard enough from ASM vendors is having APIs be part of the scope. DISCUSSION | MORE
💡 If I had one existing security space to invest in, like from companies that existed 5 years ago, I’d probably go with API Security.
The whole world of value is about to be presented as APIs, including companies. It’s core, underlying infrastructure. And they will be getting probed/tested constantly by armies of AI agents.
Yeah, API Security.
Speaking of API security and agents, my buddy Joseph Thacker wrote about this recently but I messed up the link last week. Here’s the real link, plus his new piece on a similar topic. MORE | MORE
CISA has a new alert system companies can sign up for where you sign up and get notified if you have any of their Known Exploitable Vulnerabilities (KEVs). They said about half of the companies they notified had fixed them, and that over 7,000 orgs have signed up. Super impressive. MORE
Attackers are using Microsoft Graph API for malware comms. MORE
Sponsor
Enhance Enterprise Security: Trust Every Device with 1Password!
What do you call an endpoint security product that works perfectly but makes users miserable? A failure. The old approach to endpoint security is to lock down employee devices and roll out changes through forced restarts, but it just. Doesn't. Work.
IT is miserable because they've got a mountain of support tickets, employees start using personal devices just to get their work done, and executives opt out the first time it makes them late for a meeting. You can't have a successful security implementation unless you work with end users. That's where 1Password comes in.
1Password’s user-first device trust solution notifies users as soon as it detects an issue on their device, and teaches them how to solve it without needing help from IT. That way, untrusted devices are blocked from authenticating, but users don't stay blocked.
1Password is designed for companies with Okta and it works on macOS, Windows, Linux, and mobile devices.
So if you have Okta and you're looking for a device trust solution that respects your team, visit 1Password.com/unsupervisedlearning to watch a demo and see how it works.
1Password.com/unsupervisedlearning
Watch a DemoA Russian influence campaign is exploiting college campus protests to deepen divisions in the US. The Kremlin's Doppelganger network generated over 130,000 views on X by spreading fake news about the protests, using bot accounts to mimic credible news sources. MORE
Google's making 2FA setup smoother by letting you skip the phone number for options like authenticator apps or security keys. . MORE
Sponsor
Wednesday Wisdom: Virtual Cybersecurity Showcases
Looking for a concentrated dose of cybersecurity knowledge? Join us for one (or several!) of our quick-take, 20-minute “Wednesday Wisdom” webinars. We cover timely topics around ASPM, CTEM, DevSecOps, vulnerability, and more—all led by pros in the know.
This month:
May 15: Hot Takes on Hot Topics from RSA
May 22: Your Checklist for Application Security Posture Management Buy-In.
US Marines are testing robot dogs with AI-aimed rifles. These robotic "dogs" can autonomously detect and track targets, yet require a human operator to make the final decision to fire. For now. MORE
Nearly 95% of international data travels through undersea cables, and we’re starting to see more attacks on them. MORE
💡I’ve always wondered why these weren’t a huge terrorism target. Seems like most anyone can basically turn off lots of the internet whenever they want.
Like, forget the US border—how are you going to secure cables that span thousands of miles?
I think the only real answer is to focus on the threat actors rather than the vulnerability.
The US just stopped Intel and Qualcomm's ability to ship certain goods to Huawei, right after Huawei launched an AI computer with an Intel chip. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
TECHNOLOGYA VC from Andreessen Horowitz reckons half of Google's staff are just pretending to work. David Ulevitch's claims a concerning trend of "fake work" within tech giants. MORE
💡It’s actually fake in multiple ways. It’s fake in the sense that people aren’t actually working, but it’s also fake in the sense that a lot of that work shouldn’t even exist.
I highly suggest David Graeber’s Bullshit Jobs. So good.
In this frame, AI is about to be the sunlight that’s needed to disinfect a very nasty surface. Unfortunately, a whole lot of societal infrastructure is in that filth.
The vast majority of the jobs that universities are training for simply won’t be there anymore. Middle management. Paper pushing. Spreadsheet management. Lots of project management. Customer service. Cold calling. The list continues.
Millions and millions of jobs.
Like I’ve said before other places, the safest place in this new world is building new things. Which means you need to be highly motivated and broad-spectrum in terms of your skills.
In short, the easiest way to have a job in the world that’s coming is to create something people want or need.
MITRE is partnering with Nvidia to create a $20 million AI supercomputer aimed at making U.S. government operations, from Medicare to taxes, more efficient. Simultaneously cool and terrifying. MORE
President Joe Biden is converting the Foxconn flop in Wisconsin into a $3.3 billion Microsoft AI data center. MORE
The "Acquired" podcast has become a staple for a lot of business/tech people in Silicon Valley. It basically looks at one company per episode. Kind of like the podcasts that do biographies. Super compelling. MORE
👉 Continue reading online to avoid the email cutoff issue 👈
HUMANSBiden's quadrupling tariffs on Chinese EVs, making it super tough for their cheaper cars to hit the US market. MORE
California's about to change how you're billed for electricity, introducing a fixed fee that varies with your income starting in 2025. This shift aims to lower the overall cost of electricity, with reductions between 8% and 18%, but introduces a monthly charge regardless of consumption. MORE
Scientists have found all DNA and RNA bases in meteorites, hinting life's building blocks might be extraterrestrial. This discovery, made using a novel extraction method akin to cold brewing, challenges the notion that life's ingredients originated solely on Earth. MORE
Scientists have reconstructed a 1mm square piece of the human cerebral cortex at nanoscale resolution, a breakthrough in brain research. This reconstruction allows us to see the brain's complexity like never before, offering insights into how neural networks connect and function. MORE
Vaccination has prevented 154 million deaths, according to a new landmark study. MORE
Streaming is cable now. MORE
The Emotional Support Animal Racket MORE
👉 Continue reading online to avoid the email cutoff issue 👈
IDEAS & ANALYSISJevons’ Paradox Misunderstood
Marc Andreessen was on a podcast recently and I think he mischaracterized Jevons’ Paradox as it would apply to software security. He said AI would make it harder to compete as a new business, and I think it’s the opposite.
I could be wrong here, but I don't think this is a correct application of Jevons' Paradox.
Jevons' Paradox deals with how much people use a limited resource, such as coal. The expectation is that if you make it more efficient they'll use it less, but they just use more.
1/n
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ 🧠📚✍️🗣️👥 (@DanielMiessler)
May 13, 2024
Courage is Everything
An idea I’ve been throwing around recently:
Courage is Action vs. fear
Discipline is Courage vs. laziness
Success is Discipline vs. mediocrity
Courage —> Discipline —> Success
So everything you want is on the other side of courage, whether that’s courage against fear or courage against laziness. And this applies to all sorts of real-life situations:
Hard relationship conversations
Becoming fit
Not wasting so much time with games/TV
Eating right
Quitting the soul-crushing job to pursue your calling
In this frame, it’s all Courage.
RECOMMENDATION OF THE WEEKKnow Yourself
Think about what someone should say when they introduce you.
This is _________, he __________________.
Someone introducing you
What is that sentence for you?
Mine is something like,
This is Daniel, he has a company that builds products and services that help people transition to what he calls Human 3.0 so they can survive what’s happening in AI.
The exact sentence is different for different audiences and contexts, but it’s more crucial than ever that you are able to articulate your mission to others. And that your broadcast is strong and clear enough that others can do it for you.
APHORISM OF THE WEEKThank you for reading.
UL is a personal and strange combination of security, AI, tech, and lots of content about human meaning and flourishing. And because it’s so diverse, it’s harder for it to go as viral as something more niche.
So—if you know someone weird like us—please share the newsletter with them. 🫶
Share UL with someone like us…Happy to be sharing the planet with you,
Powered by beehiiv
May 12, 2024
One Apple Fanboy's White-hot Anger at the iPad Commercial
Before I go into this, I want to articulate exactly how much of an Apple fanboy I am.
I camped for the first iPhone
I have camped for every single one since
I also camp for all other major releases, like new watches, etc.
I’ve been writing about Apple and how it’s better than its competition since 2006
I worked there for three years and it was wonderful
My whole home-tech ecosystem is Apple. I probably own 30-something of their devices, and have given away at least triple that. I spread the religion.
Anyway. Fanboy.
And that’s why I’m so fucking mad right now.
The adIf you haven’t seen it it’s basically this visually impressive crushing scene as we’ve seen before in various other places. Where you have a giant compressor and can you put things inside of it and you smash them.
They decided to put a whole bunch of creative instruments inside instead. So we’re talking like:
Pianos
Guitars
Record Players
Video Games
Paint
Books
Fucking books
You know, the things that humans actually use to create art.
And then they proceeded to fucking smash them into smithereens.
Did I mention books? They crushed books.
I actually just figured out what bothers me so much about the whole thing.
It’s like opening an ad to a violent snuff film of a whole family being killed on camera. Graphically. In detail.
Then at the end—as the punchline—they fade to black and put text on the screen.
WE PREVENT THIS
Oh, fantastic. Like the message is strong enough to counter what you just put me through.
Too fucking late guys. You just made me watch that.
You didn’t compress the instruments, and somehow put all their goodness into the iPad. That would have been kind of cool.
No. You fucking destroyed them.
And made me watch.
And then:
Ta-da! The iPad is the same! It’s the replacement!
No. It’s not. And fuck you for implying so.
This the opposite of what the company stands for. Which is human art. Human creativity.
The tech is a lens for focusing and magnifying that human creation, not for snuffing out the humanity and replacing it.
Jesus.
I worked at Apple. How did this get through the many levels of people required to approve something?
Cool so we destroy all this beautiful artistic stuff. But then we show an iPad that nobody will be able to tell the difference from the one from two years ago? Yeah, sounds good.
Someone who’s since been stuffed in a box and launched at the moon
I’m fucking ashamed. As an Apple person. Like I’m reaching out to my people and apologizing for us. Because they know how Apple I am.
Why did you make me do that?
No. Bad. Goddammit. Shit. Damn. Shit.
Anger.
—
NOTESIf I try to think objectively about how this could have happened, I see the two threads. First, compacting things is cool. Second, showing that they all fit in an iPad is cool. Nice. Two cool things. But they fucking missed that compactors DESTROY.
I saw a brilliant analysis online that said a commercial should never be better in reverse, which this one was. Nailed it.
No, I’m not worried Apple has lost its way. Shit like this happens. And when it does, people like me write pieces like this. It would take a trend to get me worried.
Powered by beehiiv
My OpenAI Event Predictions (May 2024)
I don’t have any insider knowledge—unless you count the Information article that just leaked about a possible Her-like assistant—but I think I have a good feeling for what’s coming.
How? Or what am I basing my hunches on? Two things:
Building actively in AI since the end of 2022, and
Stalking Watching Sam Altman’s comments very closely
What I anticipate
So here’s what I think is going to happen.
In a word—agents.
Here’s my thought process…
He’s basically sandbagging us. Which means saying not to anticipate much—that it’ll be incremental—and that it’ll slowly build up over time.
He’s been telling us to not expect great things in the short term
He keeps prepping us for incremental gains
This tells me he’s planning to under-promise and over-deliver in a way that surprises and delights
He’s also been telling us that the capabilities they’ll release won’t always come from models themselves, but often from supplemental and stacking (think D&D) capabilities. Like 2 + 1 + 7 + 1 = 1249.
That to me means the next functionality isn’t a 4.5 or 5 model release
This means to me things like: cooperation between models, additional sensors on models, additional integrations with models, etc.
Basically, ecosystem things that magnify models in extraordinary ways.
And I think agents are one of those things.
All about agents
Sam has talked a lot about agents—kind of in passing and in the same humble way that sets off my alarm bells. It makes me think he’s working hard on them.
It’s precisely the type of thing that could amaze people like the last live event but without announcing GPT-5.
So here’s what I’m thinking (some of this will be for future releases and not necessarily on Monday).
Agents Move Into the Prompt — The problem with CrewAI or AutoGen or any of those frameworks is that you need to do them separately from normal conversations you have with AI (prompts).
I think this will soon seem silly and antiquated.
Talking to agents (vocally or in writing) is THE PENULTIMATE (next to brain link) interface to AI. Prompting is explaining yourself clearly, and that’s the thing AI needs most.
So I think that we’ll soon be able to simply describe what we want and the model will figure out:
What needs to be a zero-shot to the model
What needs stored state
What needs live internet lookups
And how to combine all those into the answer
It’ll also be able to figure out how fast you need it based on the context. And if it needs additional information, it’ll just ask you.
So instead of defining agents with langchain, langgraph, AutoGen, or CrewAI, you’ll just say something human and rambling and flawed, like:
When a request comes in, validate it’s safe, and then see if it’s business related or personal, and then figure out what kind of business task it is. Once you know what kind of business task it is, if it’s a document for review then have a team of people look at it with different backgrounds. It needs to be pristine when it comes out the other side. Like spellcheck, grammar, but also that it uses the proper language for that profession. If it’s a request to do research on a company, go gather tons of data on the company, from mergers and acquisitions to financial performance, to what people are saying about the company leadership, to stock trends, whatever. That should output a super clean one-pager with all that stuff along with current data in graphs and infographics.
How we’ll soon create agent tasks via prompting
From there, the model will break that into multiple pieces—by itself:
Security check on initial input (we’ll use an agent that has 91 different prompt injection and trust/safety checks).
Categorize for business or personal
Categorize within business
Team of writer, proofreader, and editor agents specialized in different professions
Team of agents for researching the performance of companies by pulling X and Reddit conversations, Google results, Bloomberg dashboard analysis
An agent team that creates the financial report
An agent team that creates insanely beautiful infographics specialized towards finance
And it’ll figure out how big and expensive those agent teams need to be by either knowing the person asking, and the projects they’re working on, or by asking a few clarifying questions.
A personal DA
There’s a rumor that they’ll be launching what I have been calling a Digital Assistant since 2016. Which at this point everyone is talking about. The best version I’ve seen of this was from the movie Her, where Scarlett Johansen was the voice of the AI.
If that’s happening—which it looks like it is—that could be the whole event and it would also feel like a GPT-5-level announcement.
And that would definitely be agent themed as well, but I am hoping it’ll be more of what I talked about above.
A mix of agent stuffOne possibility is that they basically have an Agent-themed event.
They launch the Digital Assistant
They launch native agents in a new GPT-4 model, which allows you to create and control agents through direct instructions (prompting)
They give you the ability to call the agents in the prompt, like I’m talking about above, and like you currently can with Custom GPTs
If they go that route, I think they’ll likely sweeten the event with a couple extra goodies:
Increased context windows
Better haystack performance
Updated knowledge dates
A slight intelligence improvement of the new version of 4
SummarySo that’s my guess.
Agents.
Sam is sandbagging us in order to under-promise and over-deliver
This ensures we’ll be delighted whenever he releases something
He’s been hinting at agents for a long time now
Prompting is the natural interface to AI
Agent instructions will merge into prompts
Eventually we’re heading towards DAs, which he might start at the event
I’ll be watching this event with as much anticipation as for WWDC this year. No meetings. No work. Just watching and cheering. Can’t wait.
—
And if you’re curious about where I see this all going in the longer term, here’s my hour long video explaining the vision…
Powered by beehiiv
Daniel Miessler's Blog
- Daniel Miessler's profile
- 18 followers
