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 404 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 35 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
Franceschi: AI Token Spend Will Become Every Company's Biggest Expense
“I think the, you know, it will be the biggest expense in a company, like easily.”
Tan: $100K AI token workloads will drop to $200 in two years
“What you spend a 100,000 or a million dollars a year on now, it will be commonplace like in, in two years, right? It'll, it won't cost a 100,000 in a year. It'll cost 10,000. And the year after that, it'll be like a couple hundred bucks, right? And everyone wi…”
Tan: Centralized vs. decentralized AI trajectory will be decided within 24 months
“We basically have a choice to be made over the next, I don't think it's even that long. I think it's like 18 to 24 months. It might take five years, but there are sort of two scenarios”
Garry Tan: Tokenmaxxing will permeate all knowledge work across society
“I think now it's going to permeate every part of society. Like, every thing that we would call knowledge work could be token maxed.”
Garry Tan: Tokenmaxxing will not replace humans because human agency remains essential
“I don't think that it means that we're going to get rid of people. I think it means that people need to still supply the agency.”
Chaubard predicts running recursive reasoning models inside giant LLM latent spaces
“What you can imagine is we found mapping from token space or from vision, from pixels, Some really cool latent space where, like, things are just nicely semantically separated, and we can, you know, makes it really easy for downstream tasks to do, but now in t…”
Hassabis: Current AI paradigms will be part of final AGI architecture
“The components that you just mentioned, I'm pretty sure will be part of the final architecture for AGI. So I think they've come such a long way now and we've proven out so many things about what they can do. I can't see a world in which we will sort of realize…”
Hassabis: 50% chance AGI requires one or two new big ideas
“It could be that there's still one or two big ideas left that need to be cracked. I don't think it's more than one or two, if there are out there. And I think You know, my betting is about fifty-fifty if that's the case.”
Hassabis: Next few years of AI advances will come from RL and search
“And I think a lot of those ideas, both from AlphaGo and AlphaZero are really, really relevant to where we are with today's foundation models. And I think a lot of that is what we're going to see of the advances the next few years.”
Hassabis: Fixing AI reasoning gaps may only require one or two tweaks
“So there's just sort of huge gaps, I think, still, but it may only be one or two tweaks that are required to fix those kind of gaps, just to be clear, but I think that's pretty, pretty obvious there are there, and that's why you get this kind of jagged intelli…”
Hassabis: AI agents will deliver full commercial value in 6 to 12 months
“I haven't seen the result yet, which I would expect once this is really delivering that full value, which I think will come in the next six to 12 months.”
Hassabis: Frontier models will call specialized tools rather than subsume them
“One thing I can think see happening is Gemini, Claude, or one of these general systems making use of alpha fold, like specialized systems as tools. I don't think we're going to have it just in one giant brain because it will have too much regression in if I pu…”
Massad: By Agent 5, Replit will autonomously build and market SaaS daily
“So I think perhaps Agent Five, or maybe sooner, you should be able to tell Replit, every day build me a SaaS company, and like, try to market it, and see what works, and make me some revenue.”
Massad: AI agents that continually learn inside an organization are far away
“And so for us to be able to, like, deploy an agent inside an organization, and for that agent just continues to get better for that org itself is, is such a powerful thing, but that just still seems far away.”
Massad: The company of the future will consist of builders and salespeople
“I think the company of the future is made of builders and salespeople broadly.”
Chollet: LLM Progress in Non-Verifiable Domains Will Slow or Stall
“Progress of reasoning models and base LLMs on this type of domain is, is, you know, it's going to be very slow because the stack we're using, like the LLM stack is very, very reliant on its trained data. It's basically just operationalizing the trained data. A…”
Chollet: Future AI Won't Just Be Layered Harnesses on Base LLMs
“Future AI in a few decades it's not going to be this harness on top of a reasoning model on top of a base LLM.”
Chollet: AI built from first principles will be more efficient than human brains
“It's an implementation of fundamental principles, the fundamental principles of intelligence, which, you know, I think we can identify these principles and re-implement intelligence from scratch, from first principles, in a way it will be much more efficient t…”
Chollet: Artificial General Intelligence Will Likely Arrive Around ARC-6 or ARC-7
“My timeline to AGI, you know, if you just try to extrapolate from the current rate of progress and the amount of investment that's going into, not just the LLM stack, but also like side ideas, side bets that might work out, like, you know, India, for instance,…”
Mukund Jha: Foundation AI models will become completely commoditized
“Most of these models are going to get, get really, really commoditized, like, where all of these models will have similar behaviors they'll have, you know, price, price competitiveness between them”
Hodak: Ultrasound BCIs could enable digital Ambien or digital Adderall
“One of the things that I think will be possible with ultrasound is like a digital Ambien or like a digital Adderall. So can you like stimulate part of the brain to cause focus or sleep and things like that? Would not surprise me if that was possible. And that …”
Hodak: Restorative BCIs will eventually surpass baseline human capabilities
“My view on this is not that I think healthy, 30 year olds are going to be getting these soon, but eventually many people become patients. Aging is like the correlate of kind of everything getting worse. And so there's some critical age where it kind of crosses…”
Hodak: BCIs will achieve native 20/20 vision and color within 10 years
“I mean, I do think that you can get to close to native acuity, so kind of like your normal twenty-twenty vision. We're definitely not there yet, but I see a path to get there and be able to get Color and fill in a lot of the field of view. To be clear that it'…”
Hodak: The ultimate end of brain-computer interfaces is conscious machines
“If the end of the quest of artificial intelligence are super intelligent machines, then I think the end of the BCI quest are actually conscious machines.”
Taggar: Autonomous agents will select dev tools, spawning a parallel agent economy
“The agents are going to go out and choose
Tools to use to build things, which is going to essentially create a whole economy of agents, like picking and choosing dev tools and maybe other like products and goods and services, who knows, but it will essentially…”
Taggar: AI agents may eventually establish their own economy and currency
“The whole idea of, like, the human money versus agent money, like, that's kind of where this probably goes at some point, is, like, the, right now the agents are transacting, if you have them transact, they're going to transact in human money, because that mak…”
Friedman: Future SOTA Benchmarks May Come From Swarms of Cheaper AI Models
“It might be that like the next stuff that like is soda on benchmarks is not the most expensive newest foundation model with the most like GPU training. It's like a swarm of lower cost cheaper models working together just like humans do to solve a problem.”
Non-engineering roles at Anthropic already write code, a trend spreading everywhere
“Every single function on our team codes, like our PM's code, our designers code, our EM codes, our like everyone, our finance guy codes, like everyone on our team codes. We're going to start to see this everywhere.”
Taggar: Model commoditization sets up AI app startups for fantastic year
“Gemini surge, especially over the last few months, just feels like it returns us to a world of where exactly that, like the models are all essentially commoditizing each other. And it's just like the application layer and the startups are going to set up to ha…”
Diana Hu: AI's future Google- and Facebook-scale startups have not started yet
“Startups like the future Facebook or the future Google are yet to be started because those come in in the deployment phase, because right now I think these things, things are still getting built up.”
Skates: Amplitude is building a 'Cursor for analytics' product
“We're gonna go much, much bigger in December, January, and February where we are gonna be coming out with what we call the cursor for analytics, which I'm incredibly excited about, and I think will dramatically change how people use and leverage analytics”
Mignano: AI will eventually generate dynamic social media feeds without human creators
“Sora, to me, feels like this third, the start of this third phase where Eventually they don't really need creators to make content, right? Yes, today people are prompting, but you could very easily imagine a world in which you're just coming into the feed, and…”
Fisher: Post-AGI capital won't need labor buy-in, worsening wealth concentration
“But once AGI arrives, you don't need labor buy-in anymore. Right. Like you don't need folks like me to approve of the thing that you're building or of your morals or whatever. Right. Like capital begets capital in that world. And that can easily spiral out of …”
Anthropic's Joseph: Certain AI Alignment Pieces Will Move to Pre-Training
“I do think at some point there will be, like, some pieces of alignment that, like, you do want to export back into pre-training because that might be a way to, like, Put them in with more strength, like, more robustness, kind of, or more core to the intelligen…”
Anthropic's Joseph: Scaling alone likely will not achieve AGI without further paradigm shifts
“Like I think the sort of shift towards more RL is like one paradigm shift in the field, and I think it's, I think there will probably be more. I think a lot of people sort of argue about like, oh, it's like, you know, current paradigm's enough to get us to EGI…”
Pollak: Content and creator capital markets will become a major asset class
“I think one of the biggest asset classes that we believe is about to emerge is content and creators and Creator capital markets where for the first time you're going to have creators that start to say, Hey, I want to use this new platform to raise capital for …”
Levie: In 10 years, most enterprise work will be previously uneconomic tasks
“If we go 10 years into the future, The vast majority of work that gets done in 10 years from now will be work that today is in that category. It's the work that like right now we can't even attack because we're like, I'm not going to hire somebody, pay them a …”
Levie: 100 Startups Created Between 2024-2028 Will Reach $5B to $20B Valuations
“There will be 100 startups that get created between last year and in three years from now that all become five, 10, twenty billion dollar companies because they're able to find the next set of nouns and verbs”
Levie: Customers will pay for workflow software rather than raw tokens
“So eventually we're going to get to a point where customers are no longer going to just be paying for the intelligence tokens. They're going to be paying for the workflow software that goes on top of the tokens themselves.”
Levie: AI tool makers will maintain flat pricing despite falling compute costs
“It's like, it's 20 bucks a month, it's 50 bucks a month, but we know that in 10 years from now, they'll probably be able to drive down their raw materials lower, but we won't be able to command lower prices for those things, because it's just like within a rea…”
Levie: Incumbents will only sell AI agents to existing install bases
“Because I think incumbents are only going to be the agent providers for their existing install base, which means that there's going to be tons of opportunity for so many more agents that those incumbents aren't already selling into.”
Masad: Developers will reliably run 1,000 concurrent AI agents within two years
“Or plus, which I assume we're going to get to in the next couple of years, you can really spin up a thousand agents, give them a lot, thousand problems, and reliably be confident that like, 95% of them is going to work.”
Masad: Computer-use AI models will start automating real jobs within six months
“Right now, the edge of what's possible is like computer use, in my opinion, is really at the frontier of what these models could do, and I think over the next three to six months, they're going to get a lot better, and it's going to enable an entire new market…”
Masad: Tech-empowered sovereign individuals creating immense wealth will become the norm
“And this idea of like a sovereign individual, someone so empowered by technology, so empowered by these agents. That is able to create enormous amount of wealth individually is, is is, is going to be the norm.”
Masad: Single anonymous individuals creating trillion-dollar projects will become common
“It's a big market right now, but it was created by a single person, and we don't know who they are, and I think that's going to be a common occurrence in the future.”
Masad: Falling transaction costs will drastically reduce need for full-time employees
“As the cost of transaction goes down to zero, then the reason to hire an employee full time you'll have less of a reason to hire full-time employees.”
Masad: Next-gen coding AI will train via AlphaZero-style RL, not human code
“My bet is that Pretty soon we're going to move into more of the alpha zero style of training where you have a more traditional LM that's trained on all of the internet. But then the way to train the next generation of it would be to give it a reinforcement lea…”