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 51 resolved? a statement only gets an assessment when the public record can support or contradict it. opinions and what-ifs never can, and 1 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
Midha: Slowing frontier AI delays critical medical breakthroughs like curing cancer
“Until we've solved cancer, every month that we're not rushing to the frontier of accelerating biological discovery or scientific progress is a month that millions of people are suffering from disease that we could be solving with the eye.”
Midha: Existing regulatory tools are sufficient to manage AI risk
“Marginal risk is, I think, a slightly just more technical way to say, we have the tools to manage risk. We don't need new ones.”
Midha: Claims of a multi-year US AI lead over China were false
“No empirical evidence at the time that this was, that AI was creating net new marginal risks, and just factual inaccuracies that we were ahead of China, because if you just paid attention to what's happening, DeepSeek had already started to publish a fantastic…”
Midha: Distilling frontier capabilities from US AI outputs is easy
“No, actually it's not that hard to distill on the outputs of our labs.”
Midha: Downstream liability for open-source AI developers destroys the innovation ecosystem
“The status quo is you do not hold scientists liable for downstream uses of their technology. That's absurd. That's a great way to shut down the entire innovation ecosystem and start throwing literally like researchers in jail.”
Midha: US AI policy must prioritize the global race against adversaries
“We're in a race with adversaries, and that should be the calculus we should be working backwards from.”
Midha: Open-source and closed-source AI are distinct commercial markets
“Closed source and open source are not just differences in technology, but completely different markets altogether. They serve different types of customers. And so, and I think if you believe AI is this sort of explosive new platform shift, then there'll be win…”
Midha: Sesame's research preview is the ChatGPT moment for voice AI
“Relative to five, six days ago, I don't think it's too much of an exaggeration anymore to say this research preview has been for voice what ChatGPT was for text.”
Midha: Sovereign AI Will Be Largest Purchasing Decision in Next 2 Years
“It's the single largest probably purchasing decision That's going to happen in the next 24 months is do nation states start buying?”
Midha: Total national AI stack independence is impossible for most countries
“For most countries, it's impossible to have total infrastructure independence at all parts of the stack. What is much more feasible is to be great at one part of the stack and then collaborate with another sovereign or another country or region to achieve join…”
Midha: Holding model developers liable for misuse entrenches big tech incumbents
“Essentially forces most startups to lose much needed ground to big tech companies, and that entrenches incumbents more.”
Midha: AI interpretability transitioned from research to an engineering problem
“The first is that That interpretability is now an engineering problem as opposed to an open-ended research problem.”
Midha: AI regulatory burden of proof rests on those alleging extraordinary risk
“Extraordinary claims require extraordinary evidence. And so the burden of proof should be on the party making the extraordinary claims.”
Midha: AI models should be viewed as biological systems, not traditional software
“You got to think about these AI systems as almost biological systems that are grown, not coded up, right?”
Midha: Real-time testing will replace static AI benchmarks like MMLU
“While benchmarks like MMLU and the idea of these static exams were useful three years ago, the future is about real-time evaluation, real-time systems, real-time testing in the wild.”
Midha: AI ecosystem will bifurcate into Western and Chinese blocs
“The way the internet worked out was there essentially ended up being two internets, right? The Chinese internet and the rest of the world. AI may not be, end up looking that different.”
Midha: Non-frontier nations must align with major AI power centers
“All the smaller folks have to figure out which of the hyper centers they want to align with. And how do you become a modern day Singapore, Ireland, Luxembourg, et cetera, for the world of AI infrastructure. It starts with deciding whether you want to be a comp…”
Midha: Developing a US alternative to ASML will take over 10 years
“Is it feasible for the U.S. To say we're gonna build our own ASML like tomorrow? No. I mean, it's just gonna take like 10 plus years, right? EUV lithography just takes a really long time.”
Midha: Only a handful of global research teams can train frontier models
“There's only a handful really of research teams globally that are capable of this.”
Midha: US states introduced over 700 AI bills in 2024
“In 2024 alone, I think there were more than 700 pieces of state level legislation that were AI specific.”
Midha: Lack of federal data framework handicaps US AI development
“And so I think one area where we are just handicapping ourselves is that there's no unified framework in the United States at the federal level yet for data. Especially around training.”
Midha: Government balance sheets drive unprecedented Nvidia purchasing orders
“So you've had an enormous amount of NVIDIA's purchasing orders come from the balance sheet of governments. Just unprecedented demand they've been seeing from nation states realizing that they want to be hyper centers.”
Midha: Researchers can now decompose neural networks into interpretable features
“The breakthrough here was that now we've learned how to decompose a neural network into these Interpretable features when previous approaches focused on interpreting single neurons.”
Midha: AI models performed poorly on Stanford physics evaluations
“I spent some time running evals on a bunch of these models at the Stanford physics lab earlier this year, and the results were that the models are terrible at scientific analysis.”
Midha: Startup, Big Tech, and academic AI interests have diverged
“Little tech is extraordinarily different from big tech, which is extraordinarily different from the academic communities.”
Midha: AI raises creative ceilings when used by master craftsmen across domains
“When you put AI in the hands of a master craftsman from one domain, and let them expand to a new domain, It raises the bar in what they can create because you're no longer trapped by the expert of that prior domain to give you more and more of the same stuff”
Midha: Most Discord voice activity is casual, low-energy hanging out
“You know, we spend a lot of time at Discord and, you know, a huge amount of Discord usages in voice channels, and when you look, when you actually kind of spend time looking at and trying to observe what the shape of a great conversation and voice is on Discor…”
Midha: AI diffusion rate is among the fastest of general-purpose technologies
“AI has already percolated throughout society at one of the fastest diffusion rates of any general purpose technology.”
Midha: Humanity Has Produced Only Around 20 General-Purpose Technologies
“In the history of humanity, we've only had maybe 20 or 22 or so general purpose technologies, like electricity, the printing press, that have very broad based Applications in society.”
Midha: Frontier AI development is shifting from tech companies to nation-states
“And now what we're seeing is a, is shift from those, just those companies driving a bunch of frontier AI to countries and regions driving it.”
Midha: AI models fundamentally encode regional cultural values
“And this is the really important thing about AI models and how they're different from infrastructure like electricity, is there's a fundamental encoding of human values in AI models, because They're trained on data. And the data has these local norms and cultu…”
Midha: AI competitiveness depends on four core ingredients
“The good news is that there's only three or four ingredients here that really matter. The first is compute, which we've talked about. The second is abundant and low cost energy, which powers the data centers. The third is data, just the availability of really …”
Midha: Allied nations will jointly train shared AI models
“One of the things we may end up seeing is jointly trained models between countries.”
Midha: China's 2017 law forces companies to share tech with state
“There's a law called the PRC, 20 17 national intelligence law that says Chinese individuals and entities are required to support PRC national intelligence work by law, which means. If there's any technology that a PRC company has access to, they are automatica…”
Midha: US private tech is default protected from mandatory state sharing
“By and large, the private sector in the United States, and most other allied countries, is by default protected from having to make its technology available to the government.”
Midha: AI models have largely avoided dual-use national security categorization
“And by and large, AI models have not been categorized as being dual use or protected under national security.”
Midha: Lack of allied data sharing hurts frontier AI research
“One of the things that, that hurts frontier research in the United States and allied countries is a lack of government support in collaborating across borders to make more data available to allied regions.”
Midha: France's nuclear investments created extraordinarily efficient data centers today
“You know, France for example's embrace of nuclear, 20 years ago, has positioned them to have extraordinarily efficient data centers today.”
Midha: AI data centers are the new atomic unit of national sovereignty
“The leading indicator is definitely compute. If you think about the AI supply chain, the first mile starts at the data center, right? That's the new atomic unit of sovereignty, I would say, which is a new thing.”
Midha: Nations place GPU orders 12 to 36 months in advance
“And that starts with them placing orders, 12 to 36 months in advance to take delivery of GPUs, because if you don't get in front of that line, it's over. You're getting it after everybody else, right?”
Midha: Uncertainty drives AI labs from open source to closed source
“There's so much uncertainty about who wins and who loses, et cetera, that the natural tendency, I, I've seen a natural tendency for people to go to start sharing less openly with the community when they have a breakthrough or results. You know, things go close…”
Midha: The AI industry cannot currently observe internal LLM decision-making
“Where we are in the industry right now is that from the outside, we can't really see what's happening in these kitchens. So you have no idea how they made that decision.”
Midha: Interpretability breakthroughs enable high-level insight into AI decisions
“We can't control every individual cook. But now we can get insights into the bigger, more meaningful decisions that determine what meal the AI chooses to make.”
Midha: Existing tools for controlling AI models lack precision for critical uses
“We have very blunt tools to control these models, but nothing precise enough for those mission critical situations.”
Midha: AI's Value Has Mostly Been Created in Low-Precision Consumer Uses
“And so low precision environments, Consumer use cases where people are more forgiving and tolerant of mistakes by the model and so on is largely where the bulk of the value has been generated in, in, in AI today.”
Midha: LMArena serves as humanity's real-time exam for AI models
“It sounds like what Irina is, is humanity's real-time exam.”