The Ledger
Every statement that passed quotation and attribution checks. Mix any filter with any other: certainty 1/5, debate potential 5/5, or both at once.
why aren't all 5,759 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 49 checkable ones are still open, waiting for their date. predictions held up or didn't; assertions are supported or contradicted. on every card: ▮▮▮▮▮ certainty · ▮▮▮▮▮ debate potential. speakers are clickable
Ng: The biggest AI economic opportunities must be at the application layer
“And even though a lot of the PR excitement and hype has been on these technology layers, it turns out that almost by definition, the biggest opportunities have to be at the application layer. Because we actually need the applications to generate even more reve…”
Ng: Agentic AI is the most important technology trend in artificial intelligence
“And in terms of AI tech trends, if you ask me what's the most important tech trend in AI, I would say Is the rise of agentic AI.”
Ng: Subject matter expert intuition beats slow data for startup decisions
“Getting data for a lot of startups is a slow mechanism for making decisions. And a subject matter expert with a good gut is often a much better mechanism for making a speedy decision.”
Ng: Code itself is becoming much less valuable as engineering costs drop
“One surprising thing is, we're used to thinking of code as this really valuable artifact because it's so hard to create, but because the cost of software engineering is going down, code is much less of a valuable artifact as it used to.”
Ng's teams rebuilt a codebase three times in one month using AI
“So I'm on teams where, you know, we've completely rebuilt the code base three times in the last month, right? Because it's not that hard anymore to just completely rebuild the code base, pick a new data schemer, it's fine, because the cost of doing that has pl…”
Ng: AI turns software architecture decisions into reversible two-way doors
“So choosing the software architecture of your tech stack used to be a one-way door. Once you're built on top of a certain tech stack, you know, set a database schema, really hard to change it. So that used to be a one-way door. I don't want to say it's totally…”
Ng: Product management is becoming the software bottleneck as engineering accelerates
“With software engineering becoming much faster, the other interesting dynamic I'm seeing is that the product management work, getting user feedback, deciding what features to build, that is increasingly the bottleneck.”
Ng: One project team proposed staffing twice as many PMs as engineers
“So literally yesterday, one of my teams came to me, and for the first time, when we're planning a head comfort project, this team proposed to me not to have one PM to four engineers, but to have one PM to 0.5 engineers. So the team actually proposed to me. I s…”
Ng: New foundation models will not casually wipe out thousands of startups
“We're so powerful that by training a new model, we will casually wipe out thousands of startups. That's just not true. Yes, Jasper ran into trouble, small number of companies got wiped out, but it's not that easy to casually wipe out Thousands of startups.”
Ng: AI safety is a function of the application, not the technology
“Safety is not a function of technology. It's a function of how we apply it. So like electric motor, you know, You can't, the maker of electric motor can't guarantee that no one will ever use it for an unsafe downstream task. Like using electric, electric motor…”
Garry Tan: CS majors face higher unemployment than art history majors
“The craziest stat, I think this came out of the New York fed in February of this year computer science majors You know, obviously this is not the people in this room. This is just, like, out of, like, you know, a normal distribution of all computer science maj…”
Hu: Cursor scaled from $1M to $100M revenue in one year
“We've seen crazy growth unlike anything, only possible with, right now with AI, like all these companies that we work with, zero to twelve million in 12 months. The cursor story where they went zero to one in one year, the next one to a hundred. This is, like,…”
Taggar: College students often extract better AI model performance than PhDs
“The technical expertise is now actually really like the missing piece for a lot of these things. And we consistently see, at least in YC, that college students are actually at the forefront of this stuff, like actually understanding how to use the models and h…”
Chollet: Scaling LLMs 50,000x Only Lifted ARC Accuracy to 10%
“And from at the time, back in 2019 to now, with a model like GPT 4.5, for instance, there's been a roughly 50,000 X scale up of basal alarms. And we went from zero percent accuracy on that benchmark to roughly 10%, which is not a lot.”
Chollet: Displaying skill across tasks does not demonstrate intelligence
“Intelligence is a process, and skill is the output of that process. The skill itself is not intelligence, and displaying skill at any number of tasks does not show intelligence.”
Chollet: Human exam benchmarks cannot measure progress toward AGI
“And that's the reason why using exam-like benchmarks with AI models is a bad idea. They're not going to tell you how close we are to AI. Because human exams weren't designed to measure intelligence. They were designed to measure task-specific skill and knowled…”
Chollet: Base LLMs score 0% and static reasoning scores 1-2% on ARC-2
“Well, if you take Bazel Alums, model Slack, GPT-IV-IV-V, LAMA-IV, it's simple, they get zero percent. There is simply no way to do these tasks simply via memorization. Next, if you look at static reasoning systems, so systems that use a single chain of tasks t…”
Chollet: Gradient descent requires 3 to 4 orders of magnitude more data than humans
“Gradient descent requires vast amounts of data to distill simple abstractions. Many orders of magnitude more data than what humans need. Roughly three to four orders of magnitude more.”
Chollet: SOTA test-time adaptation takes thousands in compute to solve ARC-1
“Even the latest set of the art CTA techniques they still need thousands of dollars of compute to solve arc one at human level. And that doesn't even scale to arc two.”
Chollet: AI cannot reach high capability without combining Type 1 and Type 2 systems
“I really don't think that you're going to go very far if you go all in on just one of them, like all in on type one or all in on type two. I think that if you want to really unlock their potential, you have to combine them together, and that's what human intel…”
Chollet: AI will evolve into meta-learners synthesizing software on the fly
“AI is going to move towards systems that are more like programmers that approach a new task by writing software for it. And when faced with a new task, your programmer like MetaLearner will synthesize on the fly a program or model that is adapted to the task.”
Fei-Fei Li: AGI will not be complete without spatial intelligence
“To me, AGI will not be complete. Without spatial intelligence.”
Fei-Fei Li: World models require structured priors beyond brute-force scaling
“Constructing world model might be a little more nuanced. The world is more structured. There might be signals that we need to use to guide it. You can call it in the shape of prior. You can call it supervision in your data, whatever it is.”
Fei-Fei Li: Modern AGI fundamentally mirrors the original 1950s AI goal
“The founding fathers of AI who came together in 1956 in Dartmouth, you know, the John McCarthy and Marvin Minsky of them, they wanted to solve the problem of machines that can think, and that's a problem that Turing, Alan Turing, also put forward a few years e…”
Green: AI shifts tech paradigm from attention to relationships and affection
“The way I kind of like simply think about it is, is that for the last two cycles, let's call it the internet and mobile, we've moved from outcomes and attention into an area of relationships and affection. And I think that's much more dynamic and it's much mor…”
Nadella: AI energy consumption requires social permission and real economic surplus
“If there's one lesson history has taught us is that if you're going to use energy, you better have social permission to use energy. So that means you've got to make sure that the output of this AI is socially useful. In other words, if we really are not creati…”
Nadella: Change management is the real rate limiter for AI deployment
“So that change management is the real rate remitter, right? Because you're now taking the means of production in an insurance company, in a financial services company, in a healthcare company, in a software company. And saying, we are going to change everythin…”
Nadella: "Artificial intelligence" is a misnomer and AI should not be anthropomorphized
“To me, this artificial intelligence is unfortunately the worst name we could have ever picked And so I'm not into this anthropomorphizing AI at all. I mean, I think of it more. I come at it as it's a tool. It's not trying to replicate how we think.”
Nadella: Just-in-time generated and packaged software will coexist
“So I think, yes, you can generate applications just in time. You could have a prefab application that is really helping with the feedback loop to the model, and I think both of these things will exist together.”
Nadella: Software engineers will become architects as everyone builds software
“What I think will happen is all of us are going to be creating software, but there is going to be a job called a software engineer. It's going to be different, but I look at it, right? You are really taking a software engineer and saying you're now a software …”
Nadella: AI model capability is overhyped relative to real-world impact
“That is the under-hyped story, I think. Because right now the over-hyped thing is the model capability, and the model capability is fantastic, but man, if we can somehow get the world to recognize that this is making a real difference in the lives of people ev…”
Nadella: Microsoft achieved physics breakthrough to fabricate Majorana quantum chip
“Finally we've had a physics breakthrough, and we were able to actually fabricate that particle, and so therefore that's what has led to this chip.”
Altman: OpenAI has an open-source model releasing soon
“We have an open source model coming out soon.”
Altman: AI companion value lies in background persistence, not hardware
“I think the key of her is not the little piece of hardware. It's that this thing got to a point where it could run in the background and feel like a sort of AI companion.”
Altman: OpenAI will soon run world's largest, most expensive infrastructure
“I think we will run the, like, largest and most expensive piece of infrastructure in the world pretty soon.”
Altman: ChatGPT will become third biggest website, possibly first
“It'll be the third at some point, hopefully someday the first if our current growth rates continue”
Altman: AI models like o3 can act as junior employees executing multi-hour tasks
“But now you start to see things where you can like really give a task to Codex, for example or to deep research, and you have this thing go off and do a bunch of stuff and come back to you with like a proposal. It's like a very junior employee that can work on…”
Altman: Startups almost always win when industry clock cycles shift dramatically
“When the clock cycle of the industry changes this much, startups almost always win, and we've probably never seen it change this much.”
Altman: Energy will eventually be the fundamental limiter on intelligence scaling
“I cannot recall ever thinking until like after starting OpenAI that they were going to be so obviously related. That, you know, the, that energy would be the eventually the fundamental limiter on how much intelligence we could have.”
Altman: AI progressed to PhD-level intelligence across most areas in five years
“In five years we have gone from this, like, thing that could barely write a sentence to a thing that is like, you know, PhD level intelligence in most areas.”
Altman predicts AI will maintain its rapid rate of progress for five more years
“Five more years I think we'll be able to maintain the same rate of progress”
Altman: AI's economic impact will be faster and steeper than the transistor's
“The analogy I like most for AI is the transistor, like the historical tech, technical analogy. You know, some people figured out like a new, really important scientific discovery. And society, the economy, whatever you want to call it, just got to work. Just d…”
Musk: AI hype and early valuations are eye-watering, though some will pan out
“The hype level on AI is, is, is pretty intense as you've seen. You know, you see companies that are, I don't know, less than a year old getting sometimes billion dollar or multibillion dollar valuations. Which I guess could pan out and probably will pan out in…”
Musk: AI could expand the economy thousands or millions of times
“You know, the economy, assuming we don't, things don't go awry, and like AI doesn't kill us all and itself then you'll see ultimately an economy that is not, not 10 times more than the current economy. Ultimately, like if we become, say, or whatever our future…”
Musk: xAI deploying 140K GB200s and 150K H200s in Memphis
“So now we've got 150,000 H 150 K H 200 and 30 K GB 200 in the Memphis training center. And we're about to bring a 110,000 GB 200 online at a second data center also in the Memphis area.”