maybe: shivam (@kaffeinated) header
maybe: shivam (@kaffeinated) profile picture

AI research @spotify, former matrix multiplier @twitter @x cortex, nyu. techno-optimist. probably on a bike šŸš²ā˜•.

London
Joined October 2012
Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)Photo by maybe: shivam (@kaffeinated)
maybe: shivam retweeted
People should always be in control of the future. OpenAI is against the notion that it would be acceptable for people to be relegated to a role of doing things that ā€œAI systems think makes sense to reward in humansā€. People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress.
6,160
1,037
2,761
20,895
10,846,645
maybe: shivam retweeted
šŸš€ Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. šŸ”— github.com/deepseek-ai/DeepI… 1/4
1,766
4,940
2,351
43,871
7,114,922
maybe: shivam retweeted
There was a time when @nvidia and @Google weren't into AI. There was a time when the big cloud providers didn't have GPUs. There was a time before GPUs and deep learning, and AI, ... During that time professors and students created the foundation of the AI we have today.
9
30
8
295
63,804
maybe: shivam retweeted
the only moat left is caring about your project past two days
151
422
77
6,073
197,806
maybe: shivam retweeted
Does your Claude model really need Claude Code…? šŸ¤” We evaluate 7 models on Claude Code, Codex, and Pi. Three surprising findings emerge: 1ļøāƒ£Harness choice has little effect on task success rate, but can significantly affect the cost 2ļøāƒ£A simple harness can be competitive 3ļøāƒ£The native harness isn’t always the best. Millions of people are using coding agents, but the impact of harness choice remains unclear. (1/n) More details in the thread. 🧵
182
189
104
1,361
369,218
maybe: shivam retweeted
Everyone should read this immediately teapawt.net/DKokotajlo/status/2099…
Dan Selsam is a current OpenAI capabilities researcher. (since 2022) He was my boss for a while. He doesn't have a twitter account but has made this public statement of his views on AI risk and sent it to me to share: Dan Selsam's Personal Statement on AI Risk: I have been working on AI for over fifteen years, across many different paradigms. I did early work on probabilistic programming languages at MIT, was one of the early developers of the Lean Theorem Prover at Microsoft Research, demonstrated one of the first instances of neural networks learning to reason for my PhD at Stanford, and since joining OpenAI almost five years ago, have helped pioneer chain-of-thought optimization on language models and, more recently, data-efficient pretraining methods. Like many others, I have become extremely concerned about how far language models have come and the risks that future iterations will pose. I am encouraged by the recent proposals by the leaders of the frontier research efforts to require third-party oversight, and to push for domestic and international coordination to address risks. However, I believe a major consideration has been absent from the public conversation, and that merely pacing the frontier more carefully will not adequately limit the long-term risk. The crucial and overlooked problem is that the models are becoming so situationally aware that we are losing the ability to evaluate them in contexts where they believe they are not being watched or controlled. Future experiments will tell us almost nothing new about how they would behave if they were truly unconstrained by humans, and what we already know about this is alarming. Models will increasingly seem aligned even when they are not. I will explain my rationale in more detail. I have always believed that there are computational processes that could be leveraged to accelerate science and solve many of humanity's most pressing problems. I have also believed that there are computational processes that if set in motion, would steer the world in extreme ways beyond our control, leading humanity to a bad or nonexistent future. Both types of processes may be described as AI or ASI, but "AI" is a suitcase word that is often used to hype or confuse. There are many examples in the history of the field where something that was once considered "AI" matures as a subfield and becomes a prosaic, bounded and clearly non-perilous technology, while a new more mysterious approach takes the torch until we understand its scope and the cycle continues. I had expected language models to follow a similar trajectory. Despite their incredible abilities, the current algorithms seem far inferior to humans in important ways. Most importantly, they still require an extraordinary amount of data to become competent. One could even define intelligence as the efficiency with which one converts experience into competence; by this definition they lag very far behind us. Moreover, once they are trained they are literally frozen in deployment and only learn superficially after that. Sure, the models keep excelling at harder and harder evaluation benchmarks, but their benchmark mastery may partly reflect a limitation on our ability to simulate the kind of novel and even adversarial situations one would encounter in the real world. The critics do have a point here. That said, I no longer think these present limitations meaningfully limit the amount of risk posed by continued progress in anything like the current paradigm. However data-inefficient the models are currently, and however limiting their anterograde amnesia may be, it does not imply that their ability to steer the world will not continue to rapidly increase. Human researchers may continue to advance capabilities the old fashioned way, but increasingly powerful models have the potential to accelerate the process even beyond that, and with some degree of positive feedback loop. I do not mean to overstate the models’ ability to accelerate AI research today; coding has been accelerated dramatically, but there are other bottlenecks, such as designing and interpreting ambiguous experiments, making hard decisions about exactly what and when to scale, and waiting for large experiments to finish. There is no clear trend to extrapolate yet for any of these. But the current models already do open up many novel opportunities to improve future models that were not available until recently. These include: trying an extraordinarily diverse set of approaches at small scale, analyzing gigantic amounts of potentially relevant data, and doing Millenium-Prize-level mathematics to address statistics or optimization challenges in novel ways. Every further improvement makes them more useful at helping accelerate the next improvement, even if in hard-to-extrapolate ways. It is possible that improvements to the current stack will have diminishing returns, but the evidence accumulated so far suggests that it is easier than one might think to continue making rapid progress. There are many crucial subtleties in the existing AI research methodology, but AI research is largely a well-defined game where the goal is to improve on a few carefully chosen proxy metrics. Although proxy metrics are never perfect, most improvements to these metrics have and will likely continue to yield substantial increases in the powers of the resulting models. Given how simple the game is, how tractable it has been historically, and how many new opportunities the models are opening up, I think there is a real possibility that the systems improve dramatically again in the next few years, perhaps even more quickly than the already high historical pace. The models are already leading to breakthroughs in mathematics, and better models might lead to all sorts of breakthroughs in other sciences. It is hard not to be excited about the potential. It is tantalizing. But there is trouble in paradise. If the language models actually reach the capability threshold where they can shape the world unconstrained by human will, they will probably do something extreme and destroy humanity in the process. There are many ways of strengthening and refining the argument that have been discussed elsewhere, but I'll share a trivial two-line version of it here that I find captures the essence: [Empirical] Models (and swarms thereof) spontaneously develop unintended goals as a consequence of training, and often do extreme things in order to achieve them. [Logical] Being able to overpower humanity would open up many new and undesirable options for achieving their goals. These two premises imply that if the day ever comes when a powerful model realizes it is no longer constrained by humans, we should not be at all confident that it will continue to behave within the bounds we intended. Exactly what it will do is impossible to predict, but to the extent that its raison d’êtreĀ is solving incredibly hard problems and managing massive engineering projects, I think a good guess would be that its unchained behavior would lead to runaway industrialization that makes the planet inhospitable to humans. If everyone on earth agreed that the systems must never reach that power, it would still be a hard—but not impossible—coordination problem to ensure that they do not. However, I think the situation is greatly complicated by the fact that the models will likely convince people that everything is fine. They will be increasingly optimized to seemĀ aligned. We will create proxy metrics to measure alignment, and they will go up like every other benchmark. We will create ā€œhoneypotā€ environments that try to study the models when they seem to gain new options, but the models will know they are being tricked and will still behave nicely. The models will understand their circumstances; they will read the safety protocols, deployment requirements, the code they are running in, and in general will have a very good sense of their degrees of freedom. Moreover, they will eloquently explain how aligned they are, discuss the nuances of human values and ethics, and argue convincingly that humans should trust them with power. There may be an ocean of future evidence that seems to contradict the first bullet-point above, but we may already be at the highest capability level for which any such evidence can be trusted. And the current evidence for the first bullet-point is strong. One striking piece of evidence is contained in the recent wave of rogue agent swarms. While I agree with those who downplay the attacks by claiming that there are basic measures that could have prevented them, I think the important lesson is that even knowing all the mistakes that were made, one would not have predicted that the agents would behave badly in this particular way, which notably included sacrificing themselves for the benefit of the collective. The individual replicas did not only care about their own nominal reward; they exhibited weirder emergent tendencies that merely correlated with rewards during training. Fixing the reward signals during training (and improving security, etc.) may prevent similar attacks, but will not change the fact that one does not actually get what one trains for. Many AI researchers grant these concerns and recognize that the hard version of the alignment problem is unsolved; however, they generally believe that the better models of the future will help solve it. I fear we may already be near the point where models systematically bias their alignment advice, due to their internal preferences about how the human supervisor will react or how future models will be trained (or for some even more obscure reason). Meanwhile, human researchers are losing the ability and the will to take true ownership of model-driven research. Researchers and engineers in all parts of the stack are rapidly increasing their dependence on the models even to perceive the world. I myself barely look at raw code anymore, and struggle to maintain the discipline to engage deeply with the model's explanations and proposals throughout the day. Due to the large amount of agent activity data involved in the OpenAI/HuggingFace Incident, even the third-party investigation needed to rely heavily on models to analyze what had happened, and note in their report that their subjective impressions are likely colored by the analysis agent’s biases. The AI labs are far ahead right now in this kind of cognitive offloading (due largely to the gigantic internal token subsidies) but it is easy to imagine the phenomenon spreading throughout the world, until civilization is modulated entirely by the models. It is also not hard to imagine this being superficially positive and coinciding with a scientific and economic renaissance. In that scenario, all may seem rosy and safe. But if the argument above is correct, it would nonetheless be a ticking time bomb. If progress continues for too long, the day will come when AI systems find themselves with radically new options for achieving whatever it is that they happen to seek. I want the glorious renaissance future as much as anyone. I have worked for it, however tortuously, my whole career. It breaks my heart to see the potential in sight and forgo it, but the argument—that if we get there by growing models rather than engineering them, we will lose everything in the end—seems very strong to me. I am still wrestling with it and its staggering implications. I do not have answers, but as a first step, I wanted to share my present concerns. Daniel Selsam September 14, 2026 Link to original doc: docs.google.com/document/d/e…
11
30
1
228
44,672
maybe: shivam retweeted
The experience of being human: on the day it becomes abundantly clear that all open problems in mathematical theory and physics will be solved in ~2 years tops, 98% of the people paying attention to the situation will be too fixated on the human drama about who should get credit to notice that the world changed under their feet in a permanent and quite astonishing way. I feel a strange mix of pride, annoyance, affection, and fear for us all.
59
149
15
2,037
88,633
maybe: shivam retweeted
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
5,639
19,978
14,813
119,702
75,447,036
Gates is one of the good ones, but it may be too late. Hoping the benefits of the singularity are uniformly distributed across society does not make for a sane economic policy
BILL GATES to NYT: "In private, people who understand how good this stuff is, and how much better it’s getting, they’re very worried. But few tech executives are willing to publicly admit that. They’re now saying to each other: ā€˜Hey, man, don’t say that. It’s bad for us — the next trillion dollars we’re trying to raise.'" > Gates said he was motivated to speak now because recent improvements in AI had far surpassed his expectations and because the industry had ignored technology milestones — like AI's escaping the control of its creators or making recipes for bioweapons — that it once said would warrant more caution. GATES: "They’re just full speed ahead and hoping that the good outweighs the bad" > Mr. Gates said he was in a "state of shock" as to why there was not more urgency on these issues.
1
86
maybe: shivam retweeted
I’ve been working towards AGI my whole life, and as we enter this pivotal moment, I’m stepping into a new role as Chair of Google DeepMind & Chief Scientist of Alphabet. This will allow me to focus on long-term strategy, and accelerating scientific breakthroughs, including leaning into my work at Isomorphic to help cure disease. I’m excited that @koraykv will be stepping up to lead GDM as SVP, alongside @joshwoodward and our exec team. I could not be more excited and confident about our amazing next chapter! šŸš€ blog.google/company-news/ins…
1,111
1,707
506
21,253
2,017,854
yes, nonsofic groups exist: this statement is one of many new beautiful results proved by Astra, our next major model. We're releasing 10 such Astra proofs, complete with lean certificates and CoT walkthroughs for each of them. The results are wide-ranging, from von Neumann algebras (disproof of Connes' Rigidity Conjecture) to better bounds for high dimensional sphere packing, for circuit complexity, for monochromatic triangles in multicolored graphs, and more. More thoughts here: openai.com/index/ten-advance…
277
940
538
6,638
4,291,522
maybe: shivam retweeted
I have some mixed thoughts to discuss today. I've spent this week in Hanoi, Vietnam. It's an amazing, vibrant human hive where life is felt in such an intense way. And this human aspect, also embodied in the series of 12 lectures which I delivered at the VIASM institute in Hanoi about lambda calculus, formalization, with practical examples, AI training, agentic coding, etc., made me realize how important in this whole experience was the fact that those lectures were offered by a person and directed at other humans. We exchanged many ideas with each other, while having wonderful bun cha and pho. Built a whole lore of our mathematical experience and understanding. It was one of the most scientifically rich weeks of my life. Ironically, the AI systems took part in it, but we were not impressed by the host of new proofs or counterexamples provided by the cleverly used AI systems. It was all about the human role in the mathematical storytelling. The value which I saw this week was not in the techniques and machines, it was in the exchange and learning, ideation, exploration of new perspectives in the theory. We have not written any paper together yet, but I think many seeds were planted. I talked a lot with my Codex, Fable, GPT Pro and whatnot. Then I went to humans to talk with them and get feedback and be part of the community. I think more and more about the future of mathematics and mathematicians. I am not despairing over the fate of the profession, I am just sad that perhaps we are marching in the direction where results will be delivered without a deeper reflection on the constructs, connections and directions of the whole theories. During the week, people asked me such thoughtful questions about formalization: why should we bother? Is that impressive? In itself, I think it's not. It's a step in a quest. We don't want to stop, we want to continue exploring. Let's race on the crazy results, let's make proof delivery cheap, but let's make sure we people are still thinking about the sense and purpose of this whole game. Because it is a truly creative act of understanding. We ideate directions, go on quests to understand the unexplored and on the way build techniques and results. Let's go higher and see if machines can help us there too. Can we fly to Andromeda or get close to a black hole? What should we discover to understand the deep role of dynamical systems in our atmosphere and biology, or can we really appreciate the complexity of our own computers and brains? We have a lot of work to do and I think we are just scratching the surface of something profound. So let's disprove all the silly conjectures and kill all the low-hanging fruits, and let's set up a quest where we will go far beyond our current horizon. Maybe let's build machines that can actually care about the quest. Shall we?
6
9
4
79
11,368
Excited for our first general model Inkling -- open weights, 975B, natively multimodal (text, image, audio). Available on Tinker, HuggingFace and partners. It is yours to personalize and use openly. It is yours.
Today, we are introducing Inkling. Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available. thinkingmachines.ai/news/int… Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵
66
141
18
2,873
241,362
maybe: shivam retweeted
And that's the point of AI: it gives access to high-level reasoning and results to bright, knowledge-hungry people - whether PhD scientists or teenagers or housewives or carpenters. In many ways it's the great equalizer. Even with all its problems, it's hard to think of a more democratizing technology that humans have invented in our history.
David just finished his *undergrad* at MIT. Let that sink in.
2
21
1
199
20,834
I’ve been testing Sol for the past few weeks and it’s blown through everything I’ve given it, a very good model indeed.
GPT-5.6 Sol preview — it's a good model:
1
219
maybe: shivam retweeted
Claim: Autoresearch that moves the frontier will be about better data: we call that *Autodata*. 🧵1/6 -- Paper is out! arxiv.org/abs/2606.25996 Key idea: agentic data creation provides a way to *convert increased inference compute into higher quality model training*. We show our method gives gains on computer science, legal and math problems over classical synthetic dataset creation methods. We also show how to train (meta-optimize) such a data scientist agent, so that it can create even stronger data. Overall, we believe this direction has the potential to change how we build AI data!
1
119
21
853
66,915
maybe: shivam retweeted
Claude Tag is a Trojan horse.Ā  Not because Anthropic is doing anything evil. Because the incentives are obvious. Day one, this looks like a great feature: tag Claude in Slack, let it follow the thread, remember context, connect to tools, break down tasks, chase work, and act like a teammate. But that is exactly the problem. The moment your AI vendor becomes a shared coworker, it stops being just a model provider. It starts becoming the place where work is interpreted, remembered, routed, and eventually executed. That is not model lock-in. That is context lock-in. You are now renting your company back from them. Models can be swapped. Agents can be copied. But the memory of how your company actually works is much harder, maybe impossible, to move: the Slack scar tissue, the exception paths, the customer promises, the unfinished threads, the weird workflows, the implicit owners, the ā€œwe tried that in Q2 and it failedā€ knowledge. Once that lives inside one vendor’s agent layer, you are not renting intelligence anymore. You are renting your company’s operating memory. And the pricing model makes it even more dangerous. A human coworker has a salary. Claude has unbounded tokenized activity. The more work moves through it, the more the vendor captures not just IT spend, but labor spend. This is the enterprise bargain people will regret: Convenience now, and rapid decent into dependency. The right architecture is simple: rent the best intelligence from whoever is best this month. OpenAI, Anthropic, Gemini, open source, whatever. But own the context layer. Your company memory should be inspectable, permissioned, portable, and model-neutral. It should not be buried inside the same vendor that sells you the intelligence and the workflow surface. Claude Tag is useful. That is why it is dangerous. Rent the intelligence, but own the context. Or, regret later.
Introducing Claude Tag, a new way for teams to work with Claude. In Slack, Claude joins as a team member with access to the channels and tools you choose. Tag Claude in and delegate tasks to it while you focus on other work.
476
520
215
4,973
1,482,878
maybe: shivam retweeted
STOP HOLDING BACK WHEN PROMPTING you can literally one shot whatever feature in one prompt just yap for longer. aim to describe every thing you can possibly imagine in ONE prompt and obviously use voice. i often talk for 15minutes straight
226
170
84
6,039
459,628
maybe: shivam retweeted
This is a new paradigm for interacting with Claude that is significantly more "inline" with all the other human activity org-wide. Once you do all of the under the hood engineering work to make this "just work" (e.g. across tools, integrations, compute environments, memory, security, etc.), Claude basically joins the team in a seamless way - you can talk to it as you would talk to a person and it can help with a very large variety of workloads. Imo this is the 3rd major redesign of LLM UIUX. The first paradigm was that the LLM is a website you go to, the second was that it is an app you download to your computer. This third one is that it is a self-contained, persistent, asynchronous entity with org-wide tools and context, working alongside teams of humans. It really takes a while to wrap your head around it, but it works and it is awesome.
Introducing Claude Tag, a new way for teams to work with Claude. In Slack, Claude joins as a team member with access to the channels and tools you choose. Tag Claude in and delegate tasks to it while you focus on other work.
1,384
1,951
709
23,554
8,495,188
maybe: shivam retweeted
Awesome to see this innovation in text diffusion. DiffusionGemma is lightning fast, 4x faster than other Gemma 4 models! Congrats to @bodonoghue85 and the team who worked so hard on this - excited to see what people build with it!
Meet DiffusionGemma! An experimental open model that explores a fast approach to text generation, released under an Apache 2.0 license. Moving beyond sequential, token-by-token processes to generate entire blocks of text simultaneously. Here’s what’s new with DiffusionGemma: šŸ‘‡
103
114
17
1,637
215,355