Daniel Miessler's Blog, page 9
July 14, 2024
Dynamic Content Summaries (DSC)
Click for Full Size
I don’t think humans are going to be consuming most content directly from the source within a few years.
Instead, I think our Digital Assistants (our personal AIs) will be creating individual summaries of the content for us, which I’m calling Dynamic Content Summaries (DCSs).
So instead of you having to read and listen to and watch thousands of hours of content per day—which is impossible—your AI will be creating these summaries for you for the most important stuff.
Scenario: Book Too Long, Please SummarizeSo imagine you’ve been told that some book is just fantastic, and you know it has some nuggets in it, but you don’t want to spend the 17 hours reading it.
Your Digital Assistant can just create you a 3-minute video of all the best parts.
It’s a 3-minute video instead of 17-hour book
It’s a perfect deepfake of the author giving that summary
The summary compresses it very well, so it kept the most important parts
So now you know the high points.
But even better, you can just tell it:
Cool, I liked that. Give me a 20 minute audio version starting on my drive home tonight.
You talking to your DA named Armin
So now you have a podcast version, also in the author’s voice, but it’s 20 minutes long instead of 3.
Any format, length, and avatarThe power of this will be the complete flexibility in the content you’ll get.
And your DA (or a service your DA has hired) will be working on these constantly for you.
Text summaries
Audio summaries
Video summaries
Of any length
Performed by any personality
SummaryMy prediction is that much of the content we consume will actually be DCS’s instead of the original format.
Our DAs will know exactly how much time we have and what would work best for us, because they can see our schedule and they know our preferences.
We’ll be consuming far more content in far less time.
And of course our DA will still sometimes recommend that we do the raw version for certain content.
Powered by beehiiv
July 13, 2024
Should You Create a Personal or Business Brand?
I think every creator might need to make a core decision of whether they're doing:
A PERSONAL brand on which you post pretty much everything, including possibly having a business aspect.
A PROFESSIONAL brand where you mostly only talk about one topic.
I think a lot of people take significant damage in their content efforts by not being clear about which of these they are.
Let me give two examples of professional brands:
TL;DRSec by @Clint Gibler
Return on Security by @Mike Privette
These are brilliant brands because the output when you sign up is expected. It's niched to a very specific field and people aren't really turned off by surprises outside that field.
A personal brand would be someone like myself, where Unsupervised Learning is a container for everything I do, which also includes membership, community, products, etc. There's a theme, and that's similar to a professional brand, but it's not really the same because you might talk about anything in your feed, vs. just things in a specific lane. The reason I'm sharing this is to say the following:
I think Personal brands are like 10x harder to grow, and if you are trying to do a career around a business, you might want to go with a Business brand first, or completely.
I think Personal brands might be deeper and stickier over time, with your TRUE FANS, but you will have to grind for a decade or more to start gaining speed in most cases.
Be very careful and deliberate about your choice to do one or the other. I saw @0livia Gallucci - 2 ✨⛵ asking about this recently as well here, and we've talked about it directly too. Which is why I'm thinking through this and capturing it here.
There is also nuance about the name of the brand. If the name of the brand is your name, it's almost by default a Personal brand, so one thing to watch out for is if it's your name but you have trained your audience to expect only one thing. So if they only expect SQL Injection content from you, and you're now turning 35 and you want to branch out...well that can be difficult.
The reason it can be difficult is because of a concept called Audience Capture, where a creator finds themselves backed into a corner and only allowed to make one type of content, e.g., SQL Injection. Don't let this happen to you.
One way to avoid this is to have separate product or brand offerings that are distinct and similar to a Business brand, but that focus on that particular area. So for example, @jhaddix has Arcanum Security now, and people expect that to be infosec content. And he presumably won't be posting tons of stuff about Comics or Cosplay stuff there. But he's still more free to be his full self on X.
I have a close friend in security who's thinking about doing relationship content, and I used to think he should just merge that in with his current output, but now I think I was wrong about that. Now I think he should break that out into a separate brand--still losely under his own name of course--but not really have any cross-posted content between them. This way he Personally can be the umbrella, with the infosec stuff in one brand and the relationship in another.
Summary and recommendationsAsk yourself if your brand is fundamentally a business or fundamentally a life passion.
If it's a business, go niche. If it's a life passion, consider a Personal brand—and remember you can always do a separate product or service under that main brand.
If you ever plan on selling your brand, don’t have it be your name, or a unified Personal brand.
Remember that it’s 10x harder (I made up the multiple, but it’s somewhere between 2x and 20x harder) to grow a Personal/Unified brand than a niched business one.
Bottom line: Consider This Carefully.
It's much harder to go back once you've moved forward.
Powered by beehiiv
July 8, 2024
UL NO. 440: RAID (Real World AI Definitions)
SECURITY | AI | MEANING :: Unsupervised Learning is my continuous stream of original ideas, analysis, tooling, and mental models designed to help humans thrive in a world full of AI.
TOCNOTESHey there!
First, some terrorist funding news…
lol - The Internet
I’m re-reading Alex Hormozi’s two business books. He has my absolutely favorite business content right now, by far. THE FIRST ONE
My buddy Jakoby is an Army Veteran and brilliant security guy, and he’s struggling pretty hard right now. He’s actively trying to avoid eviction and any help would be appreciated. GIVE WHAT YOU CAN
Fabric made it to the front page of Hacker News! HACKER NEWS
Don’t forget to sign up for the next run of AUGMENTED which runs on July 26th. SIGN UP
Got to spend some in-person time with a couple fellow entrepreneurs/creators over the weekend. Invaluable time, these physical meetups. Nothing is more useful than having peers red-team your thinking, and I highly recommend setting up similar sessions with your peeps.
My friend Monica Verma is running her CISO Masterclass soon, and she also has a newsletter!
Ok, let’s get to it…
MY WORKI just finished a resource I’ve been working on for over a week called RAID, for Real-world AI Definitions. It was going to be like 3 paragraphs but I just kept adding more definitions and more detail. READ IT
RAID (Real World AI Definitions)
An evolving set of real-world AI definitions designed to be usable in discussion
danielmiessler.com/p/raid-ai-definitions
STORIESCloudflare has a new, free tool to stop bots from scraping websites for AI training data. | by Kyle Wiggers | TECHCRUNCH
Twilio says someone got phone numbers because of an unsecured Authy API endpoint, and 33 million numbers are now on the dark web. | THE VERGE
Russian experts say they've fully analyzed the structure of the American ATACMS missile, and they believe this knowledge will help them counter one of Ukraine's most effective long-range weapons. | EURASIANTIMES
Sponsor
Curious about AI's impact on security automation?
Tines co-founder and CEO Eoin Hinchy breaks it down with Patrick Gray on the Risky Biz podcast, discussing AI’s ability to speed up workflow build times and reduce barriers to entry.
They also cover security and privacy concerns - with AI in Tines, you decide when and how your workflows interact with AI.
Some highlights from the conversation:
“Tines does security automation and they do it extremely well. As a company, Tines is extremely well placed to make use of decision engines like large language models.” - Patrick
“With LLMs, there can be a spectrum. You can use the really cheap models to do the very basic, mundane work, and slowly and responsibly increase your usage as you build trust in the system.” - Eoin
tines.com/blog/tines-on-risky-biz
Hear the full interviewNorth Korea has switched its state TV broadcasts from a Chinese satellite to a Russian one, reducing the number of people who can watch. This change was discovered by South Korea’s KBS, which relies on North Korea's KCTV for insights into the country. THERECORD
Hezbollah launched over 200 rockets and drones at Israel after Israel killed a senior Hezbollah official, Muhammad Nimah Nasser. This is the largest attack along the Lebanon-Israel border so far. ALJAZEERA
Sponsor
Discover, secure and govern genAI use
Nudge Security discovers all genAI accounts ever created by anyone in your org, in minutes. No agents, browser plug-ins, or prior knowledge of an app required.
Within minutes of starting a free trial, you’ll have a full inventory of all genAI accounts and security profiles for each provider to help you vet new or unfamiliar apps.
Get your free genAI inventory today.
nudgesecurity.com/use-cases/mitigate-ai-risks
Free TrialThe U.S. intelligence community is diving into generative AI to enhance intelligence operations, using it for tasks like search, discovery, and counter-argument generation. | by Frank Konkel | NEXTGOV
Here’s some analysis saying AI can’t be funny, but I think I disagree. | by Anya Jaremko-Greenwold | THEWEEK
💡I’m starting to see lots of AI-generated memes that are pretty damn good. This is one Joseph Thacker shared on X:
I mean, it’s not a Chapelle standup, but it’s decent. And most importantly I’ve not seen a single explanation for why AI will continue improving in other areas but not in humor.
There's was a breach at OpenAI, but it looks like the hacker only accessed an employee discussion forum. | by Devin Coldewey | TECHCRUNCH
Nvidia is set to make $12 billion by selling over a million HGX H20 GPUs to China in 2024. These GPUs are designed to comply with U.S. export restrictions while still offering strong performance. | TOMSHARDWARE
Apple might soon announce a deal to bring Google's Gemini chatbot to iOS 18 and macOS Sequoia. | by Filipe Espósito | 9TO5MAC
💡So that means Apple will potentially be working with all three main AI players now, which is precisely what we would expect to see.
They announced OpenAI, mentioned Meta a while back, and now Google.
Greece is moving to a six-day working week to boost productivity, despite global trends towards shorter weeks. | THE GUARDIAN
💡I don’t want gesture control; I want actual cameras. :)
David Brooks sat down with Steve Bannon to understand his vision for the global populist movement. Bannon believes the ruling elites of the West have lost confidence in themselves and are detached from their people's lived experiences. | by David Brooks | NYT
The number of U.S. high school graduates is expected to peak in 2025 or 2026 and then decline for years, posing severe challenges to schools at all levels. Schools and colleges are closing, faculty members are being laid off, and districts are facing financial dilemmas. | by Jennifer A. Kingson | AXIOS
The U.S. dollar just hit a 38-year peak against the yen, driven by rising Treasury yields and Japan's sluggish economic growth. | ASAHI
People whose eyes dilated more performed better on tests of working memory, suggesting that pupil size is linked to how well we can process and remember information. | by Kate Graham-Shaw | SCIENTIFIC AMERICAN
The FDA approved Eli Lilly's Alzheimer's drug, donanemab (Kisunla), which slows disease progression by about a third. | by Tina Reed | AXIOS
High Work-In-Progress (WIP) is killing your business, innovation, and morale. The more tasks you juggle, the slower everything moves. High WIP means everyone is stressed and busy without significant results. | by Maarten Dalmijn | HACKER NEWS
IDEASThe AI Class Gap
Here's the separation that most concerns me with AI: We’re going to see even more separation between haves and have-nots because:
One small group will use AI to have a staff of 10,000 executive assistants, tutors, and analysts working 24/7 to run their businesses and optimize their entire lives.
Most people will either not use AI at all, or will only use it for gaming, watching media, porn, and doing random tasks.
It’ll largely be the same split as between voracious readers and everyone else today. The tools are available, but only a tiny percentage will use them.
The result will be a even more distance between the thriving and the surviving.
DISCOVERYSelf-Publishing a Tech Book — Andrew Wheeler shares his journey of self-publishing "Data Science for Crime Analysis with Python," offering insights on the process and tools he used. | by Andrew Wheeler | HACKER NEWS
MMA-AI — This guy is predicting MMA fight outcomes with AI and doing really well! | MMA-AI
Sam Parr launched SamsList, a database of CPAs, accountants, tax strategists, and bookkeeping firms. He’s already made over $32,000 in revenue in just two months. | by Sam Parr | SAMSLIST
The Illustrated Transformer — A visual and intuitive guide to understanding how transformers work in machine learning. | by Jay Alammar | JALAMMAR
ElevenLabs has a new AI Voice Isolator that removes unwanted ambient noise from content like podcasts and videos. | by ElevenLabs | VENTUREBEAT
Tao Te Ching by Ursula K. Le Guin — Ursula K. Le Guin's translation of the Tao Te Ching emphasizes the paradoxes and mysteries of the text, making it both profound and approachable. | by Nicholas Bennett | GITHUB
RECOMMENDATION OF THE WEEKCheck out Ground News. It’s a news site focused on seeing media bias in how stories are or aren’t covered by different media publications. GROUND NEWS
Ground News
See how different publications are presenting the same stories, or which sides aren’t even showing certain ones.
ground.news
APHORISM OF THE WEEKHey there,
You should become a member and stuff.
Here are some of the benefits:
Entry into the best online community I’ve ever been a part of
We’re voraciously curious, we’re constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Support My Work and Become a MemberThank you,
-Daniel
Powered by beehiiv
RAID (Real World AI Definitions)
I see a lot of definitions of different AI terms out there and I wanted to put my own thoughts into a long-form format. This is mostly for my own reference, but I hope it’ll be useful to others as well.
Table of ContentsThe One-Liner AI Definitions Table
Prompt Engineering (Prompting)
Retrieval Augmented Generation (RAG)
Artificial General Intelligence (AGI)
AGI 1 — Better, But With Significant Drawbacks
AGI 2 — Competent, But Imperfect
AGI 3 — Full Worker Replacement
Artificial Super-Intelligence (ASI)
Some of these are practical definitions, i.e., useful, general, and conversational containers that help us frame an everyday conversation. Others are more technical with specific thresholds, which are better for tracking milestones towards big jumps.
The One-Liner AI Definitions TableI really liked Hamel Husain’s AI Bullshit Knife that gave violently short definitions for a bunch of AI terms. Here’s my own, expanded take on it.
AI — Tech that does cognitive tasks that only humans could do before
Machine Learning — AI that can improve just by seeing more data
Prompting: Clearly articulate what you want from an AI
RAG: Provide context to AI that’s too big/expensive to fit in a prompt
Agent: An AI component that does more than just LLM call→respond
Chain-of-Thought: Tell the AI to walk through its thinking and steps
Zero-shot: Ask an AI to do something without any examples
Multi-shot: Ask an AI to do something and provide multiple examples
Prompt Injection: Tricking an AI into doing something bad
Jailbreaking: Bypassing security controls to get full execution ability
AGI — General AI smart enough to replace an $80K white-collar worker
ASI — General AI that’s smarter and/or more capable than any human
I think that’s a pretty clean list for thinking about the concepts. Now let’s expand on each of them.
Expanded Definitions TableWe’ll start with an expanded definition and then go into more detail and discussion.
AIThere are so many different ways to define AI, so this is likely to be one of the most controversial. I choose the “what used to only be possible with humans” route because it emphasizes how the bar continues to move not only as the tech advances, but also as people adjust their expectations. The general template for this rolling window is this:
Well, yeah, of course AI can do ___________, but it still can’t do __________ and probably never will.
(Narrator: And then that happened 7 months later)
I like this definition over a lot of the more technical ones because they’re usually so granular and specific, and it’s hard to get any large group of experts to agree on them.
Machine LearningI know there are a million technical definitions for machine learning, but back in 2017 when I started studying it the thing that floored me was very simple.
Learning from data alone.
That’s it. It’s the idea that a thing—that we created—could get smarter not from us improving its programming, but from it just seeing more data. That’s insane to me, and to me it’s still the best definition.
Prompt Engineering (Prompting)Some people think Prompt Engineering is so unique and special it needs its own curriculum in school. Others think it’s just communication, and isn’t that special at all.
I’ll take a different line and say prompt engineering is absolutely an art—and a science—because it’s more about clear thinking than the text itself.
Just like writing, the hard part isn’t the writing, but the thinking that must be done beforehand for the writing to be good.
The best Prompt Engineering is the same. It comes from deeply understanding the problem and being able to break out your instructions to the AI in a very methodical and clear way.
You can say that’s communication, which it is, but I think the most important component is clear thinking. And shoutout to our open source project Fabric that takes this whole thinking/writing thing very seriously in its crowdsourced prompts.
Retrieval Augmented Generation (RAG)It’s important to understand that RAG is a hack that solves a specific problem, i.e., that people and companies have vast amounts (gigabytes, terabytes, or petabytes) of data that they want their AI to be aware of when performing tasks. The problem is that AI can only practically handle small amounts of that data per interaction—either because of the size of the context window, or because of cost.
So the solution we’ve come up with is to use embeddings and vector databases to encode relevant information, and then to include small amounts of relevant context from that data in AI queries at runtime. Sending context-specific embeddings rather than the raw content makes the queries much faster and more efficient than if all the content itself was sent.
It’s not clear yet what the successor will be for this, but one option is to add more content directly into prompts as the context windows increase and inference costs go down.
AgentsThis one will be one of the most contested of these definitions because people are pretty religious about what they think an agent is. Some think it’s anything that does function calls. Others think it’s anything that does tool use. Others think it means live data lookups.
I think we should abstract away from those specifics a bit, because they’re so likely to change. That leaves us with a definition that means something like, “taking on more work in a way that a human helper might do”. So looking things up, calling tools, whatever.
The trick is to remember the etymology here, which is the Latin “agens”, which is “to do”, “to act”, or “to drive”. So ultimately I think the definition will evolve to being more like,
Perhaps that’ll be the definition in the 2.0 version of this guide, but for now I think AI Agent has a lower standard, which is anything that acts on behalf of the mission, i.e., something that performs multiple steps towards the final goal.
And like we said, practically, that means things like function calls, tool usage, and live data search.
Chain-of-ThoughtTo me, Chain-of-Thought is an example of what we talked about in Prompt Engineering. Namely—clear thinking. Chain-of-Thought is walking the AI through how you, um, think when you’re solving the problem yourself. I mean, the clue is in the title.
Again, I see prompting is articulated thinking, and CoT is just a way of explicitly doing that. I just natively do this now with my preferred prompt template, and don’t even think of it as CoT anymore.
Prompt InjectionPeople often confused Prompt Injection and Jailbreaking, and I think the best way to think about this (thanks to Jason Haddix for talking through this and sharing his definition) is to say that:
Prompt Injection is a METHOD
Jailbreaking is a GOAL
Or, more precisely, Prompt Injection is a method of tricking AI into doing something, and that something could be lots of things:
Getting full/admin access to the AI (Jailbreaking)
Getting the AI to take unauthorized / dangerous actions
Stealing data from backend systems
Etc.
Jailbreaking, on the other hand, is the act of trying to get to a Jailbroken State. And that jailbroken state is one in which as much security as possible is disabled and you have the ability to interact with the system (in this case an AI) will maximum possible permissions.
JailbreakingJailbreaking is a broader security term that applies mostly to operating systems, but that happens to also apply to LLMs and other types of AI systems.
Generically, Jailbreaking means getting to a security-bypassed state. So imagine that an operating system, or an AI system, is a very powerful thing, and that we want to ensure that users of such a system can only do certain things and not others.
Security is put in place to ensure they can’t interact with certain OS functions, data from other users, etc., and that ensures that the system can be used safely.
Jailbreaking is where you break out of that protection and get full control of the thing—whatever that thing is.
Artificial General Intelligence (AGI)In other words, this is AI that’s not just competent at doing a specific thing (often called narrow AI), but many different things—hence, general. So basically it’s some combination of sufficiently general and sufficiently competent.
The amounts of each, well…that’s where most of the debate is. And that’s why I’ve settled on the definition above. Here are a few more technical details that should be mentioned as necessary for my definition of AGI:
This AGI should be generally competent at many other cognitive tasks as well—given similar training—to roughly equal the skill level of such a knowledge worker. For example, this AGI should be able to learn spreadsheets, or how to do accounting stuff, or how to manage a budget, even if they were originally trained to do cybersecurity assessments. In other words, the general part of AGI is important here, and such a system needs to have some significant degree of flexibility.
The system should be sample efficient (learning from few examples) at about the same level as the 80K employee, recognizing that it’ll be much better and somewhat worse than a human in different areas.
The system should be able to apply new information, concepts, and skills it learns to additional new problems it hasn’t seen before, which is called generalization (again, AGI). So, if we teach it how to do budgets for a skateboarding company, we shouldn’t need to teach it to do budgets or something similar for other types of companies. So it’s not just general in its out-of-the-box capabilities but also in the new things it learns.
I like this definition because it focuses on what I would argue most humans actually care about in all this AI discussion, which is the future of humans in a world of AI. Regular people don’t care about processing speed, or agents, or model weights. What they care about is if and when any of this is going to tangibly affect them.
And that means job replacement.
So here are the levels I see within AGI—again, with the focus on replacing a decent white-collar worker.
AGI 1 — Better, But With Significant DrawbacksThis level doesn’t yet function completely like an employee, but it is right above the bar of being a worker replacement. You have to specifically give it tasks through its preferred interface using somewhat product specific language. It frequently needs to be helped back on track with tasks because it gets confused or lost. And it needs significant retooling to be given a completely different mission or goals.
Characteristics:
Interface (web/app/UI): proprietary, cumbersome
Language (natural/prompting/etc): somewhat specific to product
Confusion (focus): human refocus frequently needed
Errors (mistakes): frequent mistakes that need to be fixed by humans
Flexibility: needs to be largely retooled to do new missions or goals and given a basic plan
Discussion: A big part of the issue of this level is that real work environments are messy. There are a million tools, things are changing all the time, and if you have an AI running around creating bad documents, or saying the wrong thing at the wrong time, or oversharing something, causing security problems, etc.—then that’s a lot of extra work being added to the humans managing it.
So the trick to AGI 1 is that it needs to be right above the bar of being worth it. So it’ll likely still be kludgy, but it can’t be so much so that it’s not even worth having it.
AGI 2 — Competent, But ImperfectThis level is pretty close to a human employee in terms of not making major mistakes, but it’s still not fully integrated into the team like a human worker is. For example you can’t call it or text it like you can a human. It still sometimes needs to explicitly be told when context changes. And it still needs some help when the mission or goals change completely.
Characteristics:
Interface (web/app/UI): mostly normal employee workflows (standard enterprise apps)
Language (natural/prompting/etc): mostly human, with exceptions
Confusion (focus): it doesn’t get confused very often
Errors (mistakes): fewer mistakes, and less damaging
Flexibility: adjusts decently well to new direction from leaders, with occasional re-iteration needed
Discussion: At this level, most of the acute problems of AGI 1 have been addressed, and this AI worker is more clearly better than an average human worker from an ROI standpoint. But there are still issues. There is still some management needed that’s different/above what a human needs, such as re-establishing goals, keeping them on track, ensuring they’re not messing things up, etc.
So AGI 2 is getting closer to an ideal replacement of a human knowledge worker, but it’s not quite there.
AGI 3 — Full Worker ReplacementThis level is a full replacement for an average knowledge working in the US—before AI. So let’s say a knowledge worker making $80,000 USD in 2022. At this level, the AI system functions nearly identically to a human in terms of interaction, so you can text them, they join meetings, they send status updates, they get performance reviews, etc.
Characteristics:
Interface (web/app/UI): just like any human employee, so text, voice, video, etc., all using natural language
Language (natural/prompting/etc): exactly like any other human employee
Confusion (focus): confused the same amount (or less than) the 80K employee
Errors (mistakes): same (or fewer) mistakes as an 80K employee
Flexibility: just as flexible (or more) than an 80K employee
Discussion: At this level the AI functions pretty much exactly like a human employee, except far more consistent and with results at least as good as their human counterpart.
Artificial Super-Intelligence (ASI)This concept and definition is interesting for a number of reasons. First, it’s a threshold that sits above AGI, and people don’t even agree on that definition. Second, it has—at least as I’m defining it—a massive range. Third, it blends with AGI, because AGI really just means general + competent, which ASI will be as well.
My preferred mental model is an AI that’s smarter than John Von Neumann, who a lot of people consider the smartest person to ever live. I particularly like him as the example because Einstein and Newton were fairly limited in focus, while Von Neumann moved science forward in Game Theory, Physics, Computing, and many other fields. I.e., a brilliant generalist.
But I don’t think being smarter than any human is enough to capture the impact of ASI. It’s a necessary quality of superintelligence, but not nearly enough.
I think ASI—like AGI—should be discussed and rated within a human-centric frame, i.e., what types of things it will be able to do and how those things might affect humans and the world we live in. Here are my axes:
Primary AxesModel (abstractions of reality)
Fundamental
Quantum
Physics
Biology
Psychology
Society
Economics
etc.
Action (the actions it’s able to take)
Perceive
Understand
Improve
Solve
Create
Destroy
Secondary AxesField (human focus areas)
Physics
Biology
Engineering
Medicine
Material Science
Art
Music
etc.
Problems (known issues)
Aging
War
Famine
Disease
Hatred
Racism
Scale (things)
Quark
Atom
Molecule
Chemical
Cell
Organ
Body
Restaurant
Company
City
Country
Planet
Galaxy
etc.
So the idea is to turn these into functional phrases that convey the scope of a given AI’s capabilities, e.g.,
An AI able of curing aging by creating new chemicals that affect DNA.
An AI capable of managing a city by monitoring and adjusting all public resources in realtime.
An AI capable of taking over a country by manufacturing a drone army and new energy-based weaponry.
An AI capable of faster-than-light travel by discovering completely new physics.
Etc.
With that type of paradigm in mind, let’s define three levels.
ASI 1 — Superior➡️ Smarter and more capable than any human on many topics/tasks
➡️ Able to move progress forward in multiple scientific fields
➡️ Able to recommend novel solutions to many of our main problems
➡️ Able to copy and surpass the creativity of many top artists
➡️ Able to fully manage a large company or city itself
❌Unable to create net-new physics, materials, etc.
❌Unable to fundamentally improve itself by orders of magnitude
ASI 2 — Dominant➡️ All of ASI 1
✅ Able to completely change how we see multiple fields
✅ Able to completely solve most of our current problems
✅ Able to fully manage a country by itself
✅ Able to fundamentally improve itself by orders of magnitude
❌Unable to create net-new physics, materials, etc.
❌Unable to run an entire planet
ASI 3 — Godlike➡️ All of ASI 2
✅ Able to modify reality at a fundamental or near-fundamental level
✅ Able to manage the entire planet simultaneously
✅ Its primary concerns perhaps become sun expansion, populating the galaxy and beyond, and the heat death of the universe (reality escape)
SummaryAI terms are confusing, and it’s nice to have simple, practical versions.
It’s useful to crystalize your own definitions on paper, both for your own reference and to see if your definitions are consistent with each other.
I think AI definitions work best when they’re human-focused and practically worded. What we lose in precision can be handled and debated elsewhere, and what we gain is the ability to have everyday conversations about AI's implications for what matters—which is us.
—
NotesThanks to Jason Haddix and Joseph Thacker for reading versions of this as it was being created, and giving feedback on the definitions.
Hamel Husain has a good post on cutting through AI fluff called The AI Bullshit Knife, where he lays out his own super-short definitions. It’s quite good. X THREAD
Harrison Chase also has a model called Levels of Autonomy in LLM Applications. LANGCHAIN
I will keep these definitions updated, but I’ll put change notes down here in this section for documentation purposes.
The distinction between AGI and ASI is complex, and contested, and I’m still thinking through it. But essentially, I’m thinking of AGI in terms of human work replacement and ASI in terms of capabilities and scale.
This resource is part of what I’ve been doing since 1999, which is writing my own tutorials/definitions of things as a Feynmanian approach to learning. Basically, if you want to see if you know something, try to explain it.
Powered by beehiiv
July 1, 2024
UL NO. 439: Humans vs. AI in Prediction Markets
SECURITY | AI | MEANING :: Unsupervised Learning is my continuous stream of original ideas, analysis, tooling, and mental models designed to help humans thrive in a world full of AI.
TOCNOTESHeads-down on talks and courses and essays and video content. Feels good to have lots of ideas, but bad to be behind.
Matt Williams put out a quality introduction to Fabric on his YouTube channel. 🫶🏻YOUTUBE
—
!!! 📚Updated AUGMENTED v2 Course Outline !!!
Okay, here's an update on AUGMENTED V2—it’s now significantly better than the first version. The original plan was a big refresh, which would have been awesome, but over the last week it’s morphed into something else.
Expansion to 4+ hours vs. 3 for V1.
A whole section on augmenting AI with your personal context
A full section on building your own work/life workflows
A full section on Obsidian integration
Extensive training on—and use of—Fabric throughout
Far more examples of patterns and workflows
A more cohesive “philosophical —> technical” flow of the course
Adding more over the next two weeks…
My next couple weeks are tasked just to making this class better. Can’t wait.
🧠CHEAT CODE—become a UL member to get $250 off AUGMENTED V2.
👉 BECOME A MEMBER AND GET THE DISCOUNT
—
Ok, let’s get to it…
MY WORKI think I cracked Trump’s popularity, and unless the DNC does the same, we’re probably heading for Trump 2.
How the Far-left Will Elect Trump in November 2024
I think I finally figured out the energy source for Trump's inexplicable popularity…
danielmiessler.com/p/farleft-will-elect-trump-november-2024
STORIES🚨There's a new zero-day in OpenSSH that allows remote code execution as root on glibc-based Linux systems. It’s not super clear which SSH stacks are affected yet, but we do know that it takes some time to exploit—at least on the Morning of July 1st. Make sure you’re aware of your SSH exposure and check the resources for your stack. QUALYS
🚨There's a full 10.0 critical vulnerability in Juniper Networks routers that allows attackers to bypass authentication and take full control. | CVSS 10.0 | THEREGISTER
The Snowflake breach is expanding with over 165 victims, including Ticketek and Advance Auto Parts. A hacker from ShinyHunters claims they accessed Snowflake via third-party contractors. THEREGISTER
As part of the Snowflake incident, Santander's US branch is notifying over 12,000 employees that their personal info was compromised in a third-party breach. SECURITYWEEK
RedJuliett, a Chinese state-sponsored group, has been exploiting network edge devices to target Taiwanese government, academic, technology, and diplomatic organizations. RECORDEDFUTURE
Sponsor
Work Faster and Reduce Barriers to Automation with AI in Tines
Everyone is talking about AI right now, but the value of AI depends on your ability to trust it.
AI in Tines is designed with security and privacy in mind - you decide when and how your workflows interact with AI.
Here’s what security leaders and practitioners are saying about AI in Tines:
“Automatic mode is a gamechanger for new users, it makes the learning curve for advanced use cases very short.” - Amulya Namburi, SOAR Team Lead at Experian
"It's unlocking new use cases for us.” - Allen Cox, Senior Director of Security and IT, MyFitnessPal
“In terms of what you can build with it, the limit is your imagination.” - Kyle McGaley, Senior Security Engineer, Udemy
Try It FREERemember the R1 AI device? Well, all R1 responses ever given can be downloaded. Nightmare fuel, basically. Anything you got back from your “personal” AI device, now visible to whomever. | RABBITU
💡This is exactly what most security experts (including me) predicted with regard to AI and security. Specifically, AI startups and security.
Startups generally run with scissors, and AI startups run extra fast with extra scissors.
And if you think THIS is bad, wait until the Digital Assistant (DA) companies start popping up, and everyone is uploading their personal context to these startups. Traumas, journals, personal conversations, etc.
When those startups get breached, it’s going to be way worse.
It won’t stop DAs or AI from moving forward, mind you, because the benefits will be too powerful. But it’ll be nasty.
Here’s the AI Attack Surface as I see it:
Click for full size
Russian hacking group APT29, also known as Cozy Bear, breached TeamViewer's corporate IT environment. THERECORD
💡This is why I’ve slowly over the years migrated to a very simple stance on security tooling (or really any core tooling):
Use the official offerings from the big companies whenever possible.
Why?
They have giant security teams
They have giant security budgets
They have a LOT to lose in terms of PR and market share
In short, I only want to trust my data to companies that have both the incentive AND the resources to protect it, and those tend to be the big players like Microsoft, Google, Apple, etc.
You obviously can’t do that for all tooling, but it’s my preference when the option is available.
Chinese hackers are now using ransomware as a cover for cyberespionage, making it harder to attribute attacks. | CYBERSCOOP
Perplexity AI is under fire for allegedly stealing content from publishers and bypassing web scraping protocols. | by Cassandra Cassidy | MORNING BREW
Metaculus is launching a series of quarterly tournaments to benchmark AI forecasting against human forecasting on real-world questions, with $120,000 in prizes. | METACULUS
💡Ok this is spectacular. It’s becoming my new obsession.
RIGOROUS PREDICTIONS.
Basically, there are groups (including this Metaculus one) where people make specific predictions, and they track their success rates. So it’s not just feels; it’s real results and looking back to see how people did.
Well, THIS project is now having AI compete against humans on this!
I cannot wait to watch this field progress. NOTE: If you want to get into this world, go read SUPERFORECASTING. It’s the book that got me started in all this.
Researchers at UC Santa Cruz have figured out how to run billion-parameter-scale LLMs on just 13 watts of power, about 50 times more efficient than current data center GPUs. They achieved this by removing matrix multiplication from the LLM training and inference processes. | TOMSHARDWARE
💡This is what I call “slack in the rope”, and what Leopold Aschenbrenner calls removing “hobbling”. And it’s why I think we’re at like 1% (or way less) of our potential for neural nets and LLMs.
To me, the game is: SCALE X ALGO X TRICKS
Where “tricks” are finding this slack in the rope, which can potentially massively improve the algorithms or advantages from scale.
Leopold’s SITUATIONAL AWARENESS has the best long-form discussion of all this.
Businesses are desperate for AI guidance, and big consulting firms are stepping in to help. McKinsey says generative AI will be 40% of its business this year, and IBM's consultants have secured $1 billion in AI sales commitments. | by Matty Merritt | MORNING BREW
💡What have we been saying? 40% of McKinsey’s business!
And this stuff all started like 3 days ago, basically. 18 months is a blink of an eye. Now ask what percentage of their business is crypto-related.
Alibaba's Qwen models take three top spots on Hugging Face, while major US competitors lag behind. The new leaderboard tests models on tasks like solving 1,000-word murder mysteries and high-school math equations. | TOMSHARDWARE
💡This is disturbing. AI and drone tech are two places we need to beat China.
People in high-income democracies are increasingly dissatisfied with how democracy is working. Since 2021, satisfaction has dropped significantly in countries like Canada, Germany, Greece, South Korea, the UK, and the US. | by Pew Research Center | PEWRESEARCH
🔥 This is fine.
A study showed that loneliness in midlife is linked to believing in conspiracy theories. | NATURE
IDEASIf I designed an education curriculum, one of the main themes would be hard work → easy life, and laziness → hard life. And the concept of resilience. Honestly I would focus a lot on the Stoics, but these themes are the biggest for me. X
DISCOVERYProject Naptime — Google's new AI framework for vulnerability research lets humans "take regular naps" while it mimics human security researchers. It uses tools like a Code Browser, Python sandbox, and Debugger to improve automated discovery of vulnerabilities. | THEHACKERNEWS
💡These frameworks just keep getting better. Here’s what I think you should expect:
Remember when we saw Will Smith eating spaghetti like a year ago, and it was horrible? Well, now it looks almost realistic.
The same will happen with hacking frameworks, but I think that last 5% of skill at the top will take longer to break into for AI because it’s so heavy on human creativity.
But getting to 90-95% of what an average manual tester does in the next 2 years will be massive.
These frameworks will be used by all attackers and defenders, and the window between new vulnerabilities and either exploitation or mitigation will shorten dramatically.
Extending Burp Suite for Fun and Profit — This guide shows you how to extend Burp Suite using Montoya's methods. | by Federico Dotta | HUMANATIVASPA
ElevenLabs Text to Audio — ElevenLabs has launched an iOS app that converts any text into audio narration using AI. These voices are basically perfect, including the timing. Insanely good. | by ElevenLabs | VENTUREBEAT
Claude Projects — A new feature in Claude is Anthropic’s answer to Assistants. | ANTHROPIC
Dappier — A new platform where publishers can set a price for using their content in model training. | by Dappier | TECHCRUNCH
A Better Paradise — Absurd Ventures' new podcast "A Better Paradise" aims to elevate a fictional episodic series with a billionaire leading the world toward a digital dystopia in 2040. | by Absurd Ventures | A BETTER PARADISE
RECOMMENDATION OF THE WEEKA soon as you get a chance, go for a ride in a Waymo in San Francisco. They’re now open to everyone. It’s a remarkable experience—especially the first few times.
Try it out, and watch the screen in the vehicle. Those are all the things it’s tracking while never texting, looking at its phone, or daydreaming.
APHORISM OF THE WEEKHey there,
You should become a UL member. Here’s what you get:
Entry into the best online community I’ve ever been a part of
We’re voraciously curious, we’re constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Support My Work and Become a MemberThank you,
-Daniel
Powered by beehiiv
June 30, 2024
How the Far-left Will Elect Trump in November 2024
Click and read the full-size version before continuing
As someone who detests Trump, I think I finally figured out why he’s so popular.
In short, it’s because the far-left in the US is pushing a horribly negative view of the US that 50% of the country outright rejects.
The far-left is basically saying the following anywhere they can—including in the media and in as many school curricula as they can possibly affect.
This is what they believe—and are teaching kids to believe—in liberal areas such the San Francisco Bay Area.
The left’s mistake is not realizing how extreme this narrative is, and how little of the country agrees with it.
The whole rest of the country (75-95% ???) has a much more nuanced view of things that looks more like this:
The difference between these two narratives is why Trump is so popular.
Basically, most of the country is being forced to side with Trump because he’s the only voice sticking up for the U.S.
They don’t actually like it. They don’t like Trump. They wish there were someone better. But if it comes down to Trump or an entire narrative and party that basically hates their country, they pick Trump.
If I had to make it even more stark, it’s this:
The far left thinks the US is horrible, and that it needs to be replaced
The rest of the country thinks it’s pretty good, and worth improving
The result?
I’ve been looking for a mysterious force that’s strong enough to make someone like Trump have this much support—despite everything we know about him.
I think this is it.
That mysterious force is the forceful rejection of the narrative that:
the US is unredeemable
cops are oppressors
the rich, racism, and sexism explains everything bad happening
it’s impossible to get ahead as a regular American
anyone successful should be considered an exploiter
we should throw it all away and try communism again
Unless the Democrats realize this deep, strategic error, they might lose the White House in November. And it will absolutely be their fault.
The answer is to run someone who will push back against this horribly defeatist and anti-American take on things. Someone who will admit our flaws, defend our strengths, and demonstrate some positivity and optimism for the future.
Unfortunately, I don’t see any evidence of Democrats doing that. Biden’s administration seems captured by this sort of toxic thinking, and we are suffering because of it.
—
NOTESI could make a similar graphic for the far right vs. a logical center, and perhaps I will. It would look equally bad, with a lot of entries for attacking immigrants, rolling back women’s rights, etc., but I think this graphic matters most right now because it’s the one that’s going to get Trump elected.
Powered by beehiiv
June 24, 2024
UL NO. 438: Confusion is a Muse
👉 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 a world full of AI.
TOCNOTESRegistration for AUGMENTED v2 in July 2026 is now open. About half the slots are already gone. Sign up now by clicking below, or check UL Chat for your discount codes! RESERVE YOUR SLOT
🧠CHEAT CODE: Become a UL member to get a discount code of $250 off the course. BECOME A MEMBER
Reserve Your Slot
Fabric now supports Anthropic’s Sonnet 3.5. It’s really damn good. Just update and then run:
fabric —listmodels
echo “what’s the meaning of life “ | fabric -sp create_5_sentence_summary -m claude-3-5-sonnet-20240620
One note though, Sonnet and Haiku throw a lot of copyright failures. I guess they’re trying to push people towards Opus? Not sure why they do that, but I frequently can’t use extract_wisdom with Sonnet because it won’t summarize a podcast.
I’m gearing up for Vegas / Blackhat / DEFCON. Hope to see you there! Come see me. I’ll have stickers, but that’s about the limit of swag I’m comfortable with. I might make UL shirts at some point, but it won’t say Unsupervised Learning on them. It’ll just be the UL logo by itself on the center front, like a superhero decal.
—
🔥I’ve been feeling quite grateful for my life lately, and somehow superpowered. Not just because things are going well, which they are, but most importantly because I feel I’m living an authentic life.
I feel like this newsletter has been 300% better—at least in my eyes—because it’s lost its form and taken on the form of authenticity. It’s taken on the form of me, and what I’m thinking and doing and learning.
It’s no longer a newsletter in which I must do newsletter things. It’s now just a list of what I’m seeing and reading and learning, which I send out in something called a newsletter.
The difference is tiny and monumental, and I hope you’ve 1) noticed the difference, and 2) have enjoyed it.
Let me know. 🫶
Ok, let’s get to it…
MY WORKThis is a stream-of-consciousness style collection of thoughts on US politics, the upcoming election, and I guess politics in general. Skip it if you don’t like politics.
My View on The State of US Politics
Member Post: A set of thoughts on how I currently see US politics and the upcoming election…
danielmiessler.com/p/view-state-us-politics
Hey there!
I would like to move more to a model where I support my work with memberships, courses, and revenue from the apps that I build—rather than sponsorships.
To help me realize this, I’d like to invite you to become a UL member.
Here’s what you get:
Entry into the best online community I’ve ever been a part of
We’re voraciously curious, constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves.
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Honestly there are lots of benefits to being a member that make it worth more than the price, but at basically $8/month the community is worth it just by itself.
Become a Member🫶🏻Thank you. I truly appreciate your support of the work that I do.
-Daniel
STORIESCISA held its first AI security tabletop exercise with over 50 experts from government and industry to simulate responses to AI security incidents. The exercise, led by the Joint Cyber Defense Collaborative (JCDC), aimed to improve operational collaboration and information sharing for AI-related threats. | CISA
Extremists in the US are using AI to spread hate speech, recruit, and radicalize faster than ever. Examples: President Biden using racial slurs and Emma Watson reading Mein Kampf. | WIRED
The U.S. has banned the sale of Kaspersky antivirus software, citing security risks from Russia. If you're using it, switch to another provider. | TECHCRUNCH
There's a new set of high-severity vulnerabilities in Asus routers that let hackers take control without any user interaction. Patch these ASAP. | CVSS 9.8 | ARSTECHNICA
Sponsor
Free AI SOC Analyst for the Summer
Facing summer staffing challenges? Dropzone’s AI SOC analysts autonomously investigate your alerts, replicating the techniques of expert analysts, so your team gets to focus on only the real threats and other high-impact projects. Try it free for July and August. No playbooks or coding required - just a quick 30-minute setup. We've got your back during the busy season.
content.dropzone.ai/your-free-ai-soc-analyst-for-the-summer
Get Started in 30 MinutesDmitri Alperovitch imagines a detailed scenario of China invading Taiwan in 2028, focusing on an air assault strategy due to the rough Taiwan Strait waters. | by Dmitri Alperovitch | WIRED
The U.S. is moving to limit investments in China's AI, semiconductor, and quantum computing sectors to curb China's tech advancements. The new rules will impact private equity, venture capital funds, and U.S. limited partners' investments in foreign-managed funds and convertible debt. | TOMSHARDWARE
💡Cool, but we need to be building a lot more energy production like them as well.
Jake Lang, a January 6 rioter, has organized a nationwide armed militia from prison. Lang's militia has thousands of members and uses Telegram for coordination. | by Leah Feiger | WIRED
A study says directly at the camera during online job interviews can significantly boost your evaluation scores. | by Hiroshima University | PHYS.ORG
AI Agents and The RaaS Revolution (Results As A Service) — Discusses the evolution from SaaS to RaaS, emphasizing the role of AI agents in delivering results. MEDIUM
Sam Lijin has a new post on how to get structured output from LLMs, covering various frameworks and their pros and cons. The post dives into why converting English to JSON is tough and compares different methods like parsing malformed JSON versus constraining token generation. | by Sam Lijin | BOUNDARYML
Laura Kipnis got invited to join an AI project called Rebind, where AI versions of authors interact with readers about classic books. She thought it was a scam at first but found the idea intriguing. The project aims to make high-level, interactive reading experiences accessible to everyone, using AI to simulate conversations with authors. | by Laura Kipnis | WIRED
Fabian Both from Octomind explains why they stopped using LangChain for their AI agents, highlighting how its rigid high-level abstractions made their codebase more complex and less productive. Instead, they now use modular building blocks, which simplified their development process and increased team happiness and productivity. | by OCTOMIND | OCTOMIND BLOG
My buddy Evan and long-time ULer, Evan Oslik, argues that our obsession with convenience is making us less human, disconnecting us from real-life interactions and essential skills. | by Evan Oslik | SNAKE EYES SOFTWARE
Solar generated a fifth of global electricity at midday on the summer solstice. | EMBER-CLIMATE
Using SSH as a sudo replacement is an interesting idea. The author suggests that SSH can be used to execute commands as another user, potentially replacing sudo in some scenarios. | WHYNOTHUGO
Apple is renaming Apple ID to Apple Account. The change aims to simplify the naming but expect to see both terms used interchangeably for a while. | by Adam Engst |
A group of 17 secondary schools in Southwark, London, are going smartphone-free to combat the negative effects of phone use on students' well-being and education. | THE GUARDIAN
💡I love how Jonathan Haidt’s work is spreading so fast.
I think the vibe of “social media is the new smoking” has some merit to it.
It’s not quite that clear, though, because in moderation social media can be useful, whereas smoking is always bad for your health. Although I guess you can argue for a pro if you do it socially?
Gavin Newsom wants to ban smartphones in schools to improve student focus and reduce distractions. More impact likely from Haidt. | by Gavin Newsom | POLITICO
House prices are surging again, with a global index (excluding China) up over 3% year-on-year. American prices are up 6.5%, Australian by 5%, and Portuguese prices are soaring. | ECONOMIST
Insufficient sun exposure is now a serious public health issue, potentially causing hundreds of thousands of deaths annually in the US and Europe, and increasing the risk of various diseases like cancer, cardiovascular disease, and multiple sclerosis. The study suggests that while vitamin D has been considered the main benefit of sun exposure, other mechanisms like nitric oxide release from the skin might also play significant roles. | NCBI
💡I can’t wait for Huberman to cover this topic. Seems like getting sun early in the morning is super good for you (but that could be vibes).
My question is how much of that is circadian rhythm stuff, i.e., through the eyes, or how much is from like sun hitting the skin? And if it’s early, and the UV index is still low, does that mean it’s net-good?
Can’t wait to see more science on this. And in the meantime I’m going to treat very early and very late sun as kind of net-good.
Having more positive experiences in life is linked to lower odds of brain disorders like Alzheimer's and slower cognitive decline. A new study from Columbia suggests that brain mitochondria play a key role in this. | by Columbia Medicine | MEDICALXPRESS
Secret Ted Cruz Fundraising Documents (Slightly Redacted) — A look at Ted Cruz's fundraising events and the redaction of personal information to comply with social media policies. The documents were found in the Senate Refectory and went viral on X before being taken down for containing personal information. | by Pablo Manríquez | CAPITOL PRESS
💡One thing AI is going to bring us—and that I’m personally building—is the ability to gather intelligence continuously from thousands of sources, and then craft that into timelines and narratives.
This is the type of thing that required massive skill and headcount before. Like only journalists and spies were basically doing this, and there are/were so few of them!
Now, people like me will have AI that can gather constantly and turn that content into intelligence output.
Kolari Vision's inaugural solar eclipse photo competition crowned Ryan Spangenberg's shot of a plane skimming the eclipse as the winner. | by Matt Growcoot | PETAPIXEL
👉 Continue reading online to avoid the email cutoff issue 👈
IDEASToddler —> Ph.D in around 4 years. Insanity.
So like 3-4 years to go from toddler-level intelligence to Ph.D level intelligence.
But don't worry—it'll probably stop there.
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
Jun 21, 2024
—
Just don’t call it AI.
Click to read the thread
👉 Continue reading online to avoid the email cutoff issue 👈
DISCOVERY🔥 Web-Check — A free and open-source tool that lets you see everything about any website. You just enter a URL, click "Scan," and it gives you all the available information about the site. | by Alicia Sykes | GITHUB
Haize Labs — Automated red-teaming against LLMs. Interesting stuff going on here, will be watching closely. X
Agentic LLM Vulnerability Scanner — An open-source tool for fuzzing and stress testing LLMs with customizable rule sets. | by msoedov | GITHUB
Incogni — A service that removes your personal info from the web to block spam calls and protect privacy. | by Incogni | 9TO5MAC
Confusion is a Muse — Phil Eaton argues that confusion is a powerful tool for learning and writing, especially in software development. He says that embracing confusion and documenting the journey to understanding can lead to valuable insights and endless writing material. | NOTES
Complexity Expands to Fill the Space Available — Tom Wilson argues that complexity tends to expand to fill the space available, much like gas in a container. He suggests that limiting resources and constraints can actually drive simplicity and efficiency in systems. MEDIUM
The Winter of Content — How Game of Thrones changed media, driving traffic and homogenizing journalism. | The Verge
A niche salary database created from viral videos. | by Salary Transparent Street | SALARYTRANSPARENTSTREET
The Internet is Now for Bots, Not Humans — Chris Butler argues that the internet is now more for bots than humans, with bots making up nearly half of all internet traffic. He suggests this shift changes how we design and interact with information. | CHRISBUTLER
An impossibly thin fabric can cool you down by over 16 degrees. This fabric uses nanotechnology to reflect sunlight and dissipate body heat, making it ideal for hot climates. | FASTCOMPANY
👉 Continue reading online to avoid the email cutoff issue 👈
RECOMMENDATION OF THE WEEKGet a copy of Meditations by Marcus Aurelius and keep it by your bed.
I find it’s one of the best ways to:
Fall asleep
Feel grateful for my life as I’m doing so
Purchase: FANCY PANTS VERSION | NORMAL VERSION
APHORISM OF THE WEEKHey there,
I would like to move more to a model where I support my work with memberships, courses, and revenue from the apps that I build—rather than sponsorships.
To help me realize this, I’d like to invite you to become a UL member.
Here’s what you get:
Entry into the best online community I’ve ever been a part of.
We’re voraciously curious, we’re constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves.
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Honestly there are lots of benefits to being a member that make it worth more than the price, but at basically $8/month the community is worth it just by itself.
Become a MemberThank you. I truly appreciate your support of the work that I do.
-Daniel
Powered by beehiiv
My View on The State of US Politics
This content is reserved for premium subscribers of Unsupervised Learning Membership. To Access this and other great posts, consider upgrading to premium.
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 17, 2024
UL NO. 437: My List of Hard-won Life Lessons
👉 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 a world full of AI.
TOCNOTESI’ll be delivering my AUGMENTED course live again on July 26th!
The theme of the course this time—and going forward—will be Moving Towards Human 3.0—How to Survive and Thrive in World Full of AI. The course content will include:
HUMAN 3.0
What I think is coming for us
What I’m planning for and learning to get ready
Human 3.0, why I think it’s necessary, and how to move towards it
SKILLSETS & MENTAL FRAMES
Mandatory skillsets and mental frames to stand out after AI
What to read, study, and learn to make money post-AI
What advice to give to young people you care about
INTEGRATING AI INTO YOUR LIFE & WORK
My personal AI stack, and how I use it every day
How to build a life workflow diagram
How to use Fabric to integrate AI into your life workflow
How to create custom Fabric patterns for anything
Live Pattern-building for multiple attendee use-cases
RESOURCES
A list of my recommended books
Recommended podcast episodes
Recommended list of people to follow
Favorite AI projects and tools
Other recommended training / courses / resources
And much more…
Reserve a Limited Slot for AUGMENTED JULY 2024My friend Monica is launching her Cybersecurity Leadership Masterclass soon! She’s covering a ton of stuff, including:
How to get into a leadership position
How to build a security strategy, how to handle budget, politics, and communication as a security leader, and like a dozen other topics.
I’ve been watching the class develop and I can’t wait to see it launch. I highly recommend aspiring security leaders—or leaders who want to up their game—attend. SIGN UP FOR THE CLASS
—
Trying to make sense of AI’s role in your security operations? You’re not alone.
Join me and James Spiteri, Director of Product Management for Elastic Security, for a discussion on the benefits, risks, and considerations when implementing AI technology into your security stack.
Tuesday, June 25th, 9 AM PDT REGISTER FOR THE WEBINAR
—
I’m doing a minor modification of header titles, going from SECURITY/TECH/HUMANS to NEWS/IDEAS/DISCOVERY. The rationale is that: For me, security, tech, and humans are merging so deeply that I see them as integrated. So I think it makes more sense to break things down by new things happening, ideas that I find fascinating, and surfacing cool apps/projects/etc. If we don’t like it, I can always go back to the previous way, but I think this is more natural.
Ok, let’s get to it…
MY WORKA couple of new essays this week.
First, my new piece on how AI will soon start teaching us, testing us—and scariest of all—rating us. And how these scores will start being used everywhere…
ASTRA Scores: AI-Powered Assessment and Rating Systems
With AI, hiring, dating, and learning is about to get very strange.
danielmiessler.com/p/astra-scores-aipowered-assessment-rating-systems
And this is my new piece on how we’re replacing meaningful, slow things with shallow fast things…
The Fast-Slow Problem
We've become so obsessed with fast pleasures that we can't enjoy slow ones anymore.
danielmiessler.com/p/the-fast-slow-problem
My list of hard-won life lessons. Expanded list from what you saw on socials.
A List of My Hard-won Life Lessons
A list of things it took me a very long time to learn.
danielmiessler.com/p/list-hardwon-life-lessons
I would like to move more to a model where I support my work with memberships, courses, and revenue from the apps that I build—rather than sponsorships.
To help me realize this, I’d like to invite you to become a UL member.
Here’s what you get:
Entry into the best online community I’ve ever been a part of
We’re voraciously curious, constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves.
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Honestly there are lots of benefits to being a member that make it worth more than the price, but at basically $8/month the community is worth it just by itself.
Become a Member🫶🏻Thank you. I truly appreciate your support of the work that I do.
-Daniel
STORIESThe House just passed a bill that bans DJI drones from using FCC frequencies, and it's now moving to the Senate. This could mean no more DJI drones in the U.S. if it becomes law. Would be a trip to see this happen before TikTok. | DRONEDJ
A disgruntled ex-employee cost NCS over $600,000 by deleting all 180 of its test servers using scripts he found on Google. Deprovisioning gets so many people. I honestly think IAM will take longer to solve than AGI. That and A/V. | TOMSHARDWARE
💡I mean, there are lots of things that have to go wrong for this to happen, but we’d be lying if we said it wasn’t possible in a lot of places we’ve worked. Or still work.
Wells Fargo fired over a dozen employees for faking keyboard activity to look busy. | by Hannah Levitt | BLOOMBERG
💡So happy to see this. Physical keys are great, but I’d prefer both security and convenience. Passkeys have to be the best thing that’s happened to consumer security in 10 years.
Sponsor
Ebook: Meet Tomorrow’s IT Heroes, Today
The digital terrain of Information Technology Operations (ITOps) is experiencing a seismic shift. As we phase out old-school tech and methodologies, we usher in an era of safe automation, efficiency, and autonomous systems. This ebook is more than a glimpse into the future of IT — it’s a dynamic portal through which we explore the pioneering work of visionary professionals reshaping this landscape. Are you ready to meet the architects of IT’s future and take your place among them?
patch.automox.com/rs/923-VQX-349/images/Meet-IT-Heroes-eBook.pdf
Read It NowAWS has added FIDO2 passkeys for MFA, making accounts more secure and user-friendly. Root users must enable MFA by July 2024. | BLEEPINGCOMPUTER
New York Times source code was stolen using an exposed GitHub token. | BLEEPINGCOMPUTER
iOS 18 will let you automatically record and transcribe your phone calls through the Phone app. | by Juli Clover | MACRUMORS
Sponsor
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 TodayCanada's proposed Online Harms Act could imprison people for life over hate speech and even preemptively restrict their freedom based on anticipated crimes. | by Conor Friedersdorf | THE ATLANTIC
LinkedIn is rolling out AI career coaches that can help with everything from negotiating salaries to writing résumés. | by Amanda Hoover | WIRED
💡See my essay above on rating systems. ASTRA SCORES ESSAY
AI and LLMs are transforming cyber insurance by making real-time risk assessments, underwriting, and claims processing more efficient. | by Louis Columbus | VENTUREBEAT
💡Oh man, I haven’t done a piece on this yet, but AI is going to make all insurance companies a lot smarter about real-time risk assessments and pricing adjustments.
Scale came out with a “controversial” stance that they will hire on merit and merit alone. It’s basically an anti-DEI statement, but not in a bad way. They’re saying that it will actually lead to healthier diversity than when it’s forced by woke policies. | by Scale AI | SCALE
💡Everyone anticipated massive backlash to this statement, but there’s been very little. Which basically means the political pendulum has swung—or is still swinging—away from extreme wokeness. I just hope the pendulum stops in the middle rather than going into crazy town on the other side of this clock metaphor.
Nathan Wade's media consultant interrupted his CNN interview when he was asked about his relationship timeline with Fulton County DA Fani Willis. Wade said the relationship is a distraction and not relevant to the case. | by Lauren Sforza | THE HILL Consultant Interrupts Interview Over Relationship Timeline
Tesla shareholders have backed Elon Musk's $56 billion pay package and approved moving the firm's legal HQ to Texas. The deal was previously blocked by a Delaware judge for being unfair to shareholders. | by João da Silva and Natalie Sherman | BBC Tesla Shareholders Approve Musk's $56B Pay Package
💡This feels very Atlas Shrugged to me. Basically, give the actual builders anything they want. There are few of them left and they’re special. I like the pro-builder sentiment there, but I fear it comes with an unhealthy worship culture that breeds toxicity.
Like, “Give Buck Rhinehold a harem! Buck deserves it!” (cheering)
I’d like to see a healthy hybrid of Atlassian respect for founders/builders/creators, etc., but with the assumption that everyone should be like that, and thus, there’s no room for god-complexes.
Wow, what timing.
Elon Musk had sexual relationships with multiple SpaceX employees, including a former intern he later hired onto his executive team, according to The Wall Street Journal. Another woman alleged that Musk asked her to have his children and then denied her a raise when she refused. | by Elizabeth Lopatto | THE VERGE
💡This is so insane. I mean, I’m inspired by his willingness to just ask for what he wants. There’s a version of that where it’s healthy. But, um, not with your fucking employees.
“Julie, great work on the Montague project. I admire you; you should have my babies.” (not a real quote, obviously)
Like it’d be wrong even if nothing bad happened from it, because of the potential for it go wrong. But if he’s actually doing what they’re saying here, that’d be really bad. It’s not just making the proposition, but punishing those who refuse.
This is why it’s not allowed, people!
California has the biggest wage gap in the US. Seems to me like just the most extreme version of a big city. You’re going to have the higher highs and the lower lows. | by Tessa McLean | SFGATE
We just broke ground on America's first next-gen nuclear facility designed by TerraPower. This plant is safer, cheaper, and more reliable than existing ones, aiming to revolutionize power generation in the U.S. and globally. | by Bill Gates | GATESNOTES
China's yuan is now Russia's main foreign currency, replacing the dollar and euro. | UNN
A fungus living in the sea can break down the plastic polyethylene, but only after it's been exposed to UV light. | by Annika Vaksmaa | NIOZ
Apple left out a lot of small updates at its WWDC keynote this year. Here are some of the coolest ones: iPhone bezels now have animations, you can type in Spanglish without toggling keyboards, and the flashlight app has new animations. Plus, widgets are easier to resize, the Vision Pro shows your keyboard in VR, and macOS Sequoia has nostalgic wallpapers. | by Wes Davis | THE VERGE
Apple Vision Pro users can now watch House of the Dragon in an updated immersive Iron Throne Room environment on the visionOS Max app. | by Ryan Christoffel | 9TO5MAC
Spatial Personas in visionOS 2 can now high five, fist bump, and touch fingers with visual and audio feedback. | by Tim Hardwick | MACRUMORS
💡Excited about this. One of the things I enjoy doing with my Vision Pro is Facetime. And anything that makes it more natural is great.
Also can’t wait to try out the new dual-4K-monitor widescreen for working virtually.
Sebg argues that Jupyter Notebooks are the fast food of coding—convenient but often unhealthy for long-term projects.| by yobibyte | YOBIBYTE
Nvidia CEO Jensen Huang told Caltech grads to pursue "zero-billion-dollar markets," which are markets with no current value but huge future potential. He emphasized the importance of resilience and innovation, encouraging graduates to see setbacks as opportunities for new directions. | by Jowi Morales | Tom's Hardware
💡Just figured out what I love so much about Jensen. He’s like a nice Elon. Not saying Elon isn’t also nice; I know he is. Just saying he’s often not.
I love seeing—and having others see—that you can be a murderer-ambitious-creative-person, and still be kind most of the time.
Ro Arepally predicts that generative AI will actually increase the demand for software engineers over the next 20 years, not decrease it. The idea is that AI will make software engineering more accessible, similar to how autopilot systems made flying more accessible, leading to more tech companies and higher salaries for engineers. | by Ro Arepally | ROAREPALLY
The widely held view that sperm counts in men are dropping around the world may be wrong, according to a new study by the University of Manchester, Queen’s University in Kingston, Canada, and Cryos International, Denmark. | by UofMNews | UNIVERSITY OF MANCHESTER
Curiositry argues that perfectionism is about optimizing at the wrong scale, suggesting that focusing too much on perfecting small details can lead to missing the bigger picture. | by Curiositry | AUTODIDACTS
38% of webpages that existed in 2013 are no longer accessible a decade later. | by Athena Chapekis et al. | PEWRESEARCH
Derek Sivers explains why you should create a /now page on your website, basically a page that tells people what you're focused on right now. | by Derek Sivers | SIVERS
💡I had one of these way back in the 10’s. I’ll have one again soon, but it’ll be in my personal API. Which will also show up on my site. 😀
👉 Continue reading online to avoid the email cutoff issue 👈
IDEASAn X thread I posted about increased production of goods and services after AI, and how it doesn’t make sense if 90% of people don’t have money to buy the stuff…
There's a critical problem with moving towards AGI and ASI in a way that increasingly replaces human workers.
Who is going to buy all the stuff?
Sure—we'll be able to make 1000X more stuff, but people won't have jobs—and therefore money—so who are we making all the stuff for?
— ᴅᴀɴɪᴇʟ ᴍɪᴇssʟᴇʀ (@DanielMiessler)
Jun 16, 2024
Venkat Rao argues that in a world dominated by AI and robots, the key to human survival isn't being smarter but being more mediocre. He suggests that mediocrity is an independent meta-trait crucial for evolutionary fitness and long-term survival. | by Venkatesh Rao | RIBBONFARM
👉 Continue reading online to avoid the email cutoff issue 👈
DISCOVERYPop Culture's Oligopoly: Why Everything Feels Like a Sequel EXPERIMENTAL HISTORY
Dream Machine — An AI model that makes high-quality, realistic videos fast from text and images. | by Luma Labs AI | Dream Machine
Nvidia Warp — A Python framework for high-performance GPU simulation and graphics. | by NVIDIA | GITHUB
The "Infinite Content Ideas Generator" — A 2-step AI prompt approach to never run out of content ideas. | by Moritz Kremb | THEPROMPTWARRIOR
You can now design and manufacture your own chip, and this video shows you how. | by Robert Feranec | YOUTUBE
LLM Mojo - The project ports Andrej Karpathy's llm.c to Mojo, leveraging performance enhancements like vectorization and parallelization. | by dorjeduck | GITHUB
Ollama v0.1.45 — Enhanced GPU discovery and multi-GPU support with concurrency. | by Ollama Team | GITHUB
SQLite is likely used more than all other database engines combined. | SQLITE
Starlink Mini — A new portable Starlink dish designed for camping, RVs, and other mobile uses. | by Starlink Hardware | STARLINKHARDWARE
Max Leiter has a simple but powerful tip: ship something every day. It doesn't have to be big, just something you can point to. | by Max Leiter | MAX LEITER BLOG
👉 Continue reading online to avoid the email cutoff issue 👈
RECOMMENDATION OF THE WEEKThis week I recommend you check out my list of truths it took me a very long time to learn.
MY LIST OF HARD-WON LIFE LESSONS
APHORISM OF THE WEEKI would like to move more to a model where I support my work with memberships, courses, and revenue from the apps that I build—rather than sponsorships.
To help me realize this, I’d like to invite you to become a UL member.
Here’s what you get:
Entry into the best online community I’ve ever been a part of.
We’re voraciously curious, we’re constantly reading, constantly learning, and we share what we learn with others. But most importantly, we’re kind. We’re a community of helping people become their best selves.
An insanely great Book Club, which has been running for like 5 years straight!
Additional monthly meetups where we share tools, routines, personal challenges, and lots of other stuff you can’t get anywhere else
Significant discounts on my paid offerings
Access to private UL events
And more…
Honestly there are lots of benefits to being a member that make it worth more than the price, but at basically $8/month the community is worth it just by itself.
Become Part of Our UL CommunityThank you. I truly appreciate your support of the work that I do.
-Daniel
Powered by beehiiv
A List of My Hard-won Life Lessons
I’m working on my context.md file for my personal Digital Assistant, and one part of that will be my model.md file, which is basically life lessons, or ways I view the world.
I’ve been capturing them in various forms, but one useful way is as a set of highly-compressed pieces of wisdom.
Here’s my list of uncommon knowledge that took me far too long to discover.
In practice, weak people behave very similarly to evil people. Don’t tie yourself to either.
-
When I am feeling low mood or low energy, 95% of the time it's because I have not 1) gotten up early, 2) gotten outside, 3) worked out, 4) gone for a walk, or done anything else physical in the last 2 days. Just go do that, and you'll instantly feel better.
-
Radical honesty—or something close to it—is harder to do at the time but infinitely better for your relationships. Avoiding hard conversations creates toxicity that builds up over time and destroys things.
-
You almost never lose by being kind to a stranger. No strings attached. A compliment. A gesture. A smile. Just be nice. People really, really need it right now. Give it away freely. It feels good, too.
-
People believe things mostly for reasons they don't understand, which are largely emotional and identity-related. This is why facts and logic aren't really effective in arguments. If you want to change minds, start by figuring out why they hold their opinions.
-
Taking responsibility for the state of your life is a superpower. There are few more powerful words than, "That's on me. I will fix that.", and few more destructive words than, "I can't do that because {someone else} is doing {a thing}."
-
Meaning requires you have purpose. Purpose comes from being useful. Being useful comes from working on problems that matter. Build skills to help you work on those problems. Meaning -> Purpose -> Useful -> Problems -> Skills.
This works in the opposite direction as well. If you don't know how to get meaning in your life, think about what skills you have. Then think about what meaningful problems you can apply them to. Which will make you useful. Which will give you the meaning you seek. Skills -> Problems -> Usefulness -> Meaning.
-
I am my most optimistic and kind when I am high-energy. And I am high-energy when I work out. So working out isn't just a physical thing: it makes me a kinder, more optimistic, and overall better person who is more productive as a side benefit.
-
Procrastination and anxiety are much more related than people think. Good sleep, exercise, and general organization of life address both.
-
Your work can only be as rewarding as the problems you're working on are meaningful.
-
When I'm not exercising and I feel bad, I feel certain that exercise is not the problem. It's something else. Something worse. But as soon as I exercise, I realize the problem was that I wasn't exercising.
-
The key to creating great content is authenticity and surprise. In that order. Be yourself, constantly explore, and share that with others. It's an evergreen content strategy. Perhaps the only one.
-
The ideal number of human workers in any business is zero. The purpose of companies is to make as much money as possible with the lowest possible expenses. So AI and other types of automation are not disruptions to a human-based Capitalism—instead, they’re revealing that today’s Capitalism is not fundamentally human in the first place.
-
Few things help a relationship more than honestly asking for what you want, and both sides being willing to walk away if it's not a match. - In hiring, I used to think that everyone is curious and ambitious and talented, and you just needed to expose them to the right training. But the truth is that some people only need a task and a search engine and they're highly productive. And others can get two master's degrees and take 6 months of training and still ask you what they should do every morning.
-
Becoming comfortable with being yourself is everything. Authenticity is the only way to be truly fulfilled. You'll have some people who are jealous or who will hate; just accept it as part of the cost.
-
VCs and HR representatives aren't your friends. They're there for the company, not for you. Just because they're smiling doesn't mean they'll do right by you. The incentives are misaligned.
-
People do their best work when they're obsessed with a problem. Find problems that allow you to be like that, and hire people who feel like that about the problems you're working on. - Many people wish they could find another partner, but they stay together because of kids, bills, or family obligations that would make it very difficult to split up. Don't let yourself be one of those people.
-
Your parents are flawed people who were doing their best given their upbringing and their own traumas. Forgive them and love them as best you can.
-
Don't work in an industry; work on problems. - Don't expect someone to understand something if their job and role require that they don't. - If you have the wrong partner in life, that's your ceiling for life happiness. If they're a 4, then 4 is the happiest and most successful you can be. Your partner is the most important choice you'll make.
-
The most insane life hack in the world is somehow the most secret. Eat and exercise like a thin person and you'll be thin. Read like Charlie Munger and you'll be smart. Work out like Sam Sulek and you'll gain muscle. Copying what works actually works.
-
The way to tell who your real friends are is to think of who will celebrate a win you're having even if they're having a bad day or week or month. True friendship doesn't depend on them doing better than you.
—
NOTESI’ll keep adding more
Powered by beehiiv
Daniel Miessler's Blog
- Daniel Miessler's profile
- 18 followers
