Bharath T. (@PreemptDisease) profile picture

Founder building complex nano-bio-ML products, towards disease preemption (detect & cure before full disease onset). https://xitter.gq/t.co/ul0mf5Jqsk - scaled high-res SPR kinetic data

Menlo Park California
Joined June 2017
Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)Photo by Bharath T. (@PreemptDisease)
May be early intervention hypothesis should have been trialed in parallel, missed opportunity. Also summary proof that biomarkers of disease are ‘not necessarily’ disease targets.. correlation vs causation strikes again. Interestingly similar failures occurred in Neuro degenerative diseases
Medicine and biomedical research are very humbling. The evidence base supporting the IL-6 and Lp(a) hypotheses in atherosclerotic cardiovascular disease has been compelling. Most people would agree that these trials were worth pursuing. But no matter how much we think we understand biology, no therapy can be assumed to benefit patients until a trial shows that it does. There's clearly a lot to optimize in how we prioritize and test hypotheses. But those arguing that AI will cure all disease in 5 years should keep these results in mind.
1
22
Bharath T. retweeted
If we ban data centers, the massive investment in infrastructure (trillions) and hundreds of thousands of jobs will just flow to Canada, Asia, Eastern Europe. If we pause AI, we will become a natural resource and food supplier for a dominant China.
59
33
8
338
15,965
METR's Ryan Greenblatt estimates AI is now speeding up AI research by ***4.5x*** This is 5 MONTHS EARLIER than AI 2027 predicted. On this trajectory, we may be just months from full-blown AI takeover. "I was seriously wrong about RSI speed up." "A 2x acceleration in overall AI progress from RSI corresponds to a much larger than 2x RSI speed up for AI software R&D (algorithms and data); I did an estimate using my public preexisting modeling of drivers of AI progress and the public report which yielded a very rough estimate of 4.5x for how much RSI is currently accelerating AI software progress (relative to no AI automation). In the AI 2027 scenario, this much acceleration happens around March 2027."
This report on AI R&D automation (RSI) is a big update towards ASI within 1-2 years. It says AI progress is sped up by RSI right now by around 2x. In AI 2027, this happens around March 2027. I was seriously wrong about RSI speed up; I previously expected much less in 2026 (around 1.15x-1.3x). It seems as though there was specific privately available evidence that would have changed my mind. I haven't seen an unredacted version of this report; this is purely about the public report. Because so much of the underlying evidence was redacted from the public report, I think conclusions about things other than the observed 2x faster rate of progress on benchmarks depend significantly on trust for the authors of the report. (E.g., conclusions about whether this speed up is from RSI depend on trust.) I do trust the authors of the report a lot to be careful and rigorous and generally do a good job. (Note that I also work at METR.) Less extensive redactions would greatly reduce this need for trust. Since in my current role I may have access to other confidential information that could be relevant to timelines one way or another, I unfortunately cannot state my all-things-considered views on ASI timelines or the probability of a software intelligence explosion right now. Here, I’ve done some analysis of the implications using my previous public statements. A 2x acceleration in overall AI progress from RSI corresponds to a much larger than 2x RSI speed up for AI software R&D (algorithms and data); I did an estimate using my public preexisting modeling of drivers of AI progress and the public report which yielded a very rough estimate of 4.5x for how much RSI is currently accelerating AI software progress (relative to no AI automation). In the AI 2027 scenario, this much acceleration happens around March 2027. Here are the worked details of this estimate based on my public modeling. There are two main reasons RSI speed up of AI software R&D must be larger: - Not all of AI progress is from better software (some is from scaling up compute in training runs). I've previously speculated that around 2/3 of AI progress is from software improvements which implies that 2x overall AI progress acceleration requires 3x faster software progress. - Diminishing returns: A fixed amount of progress requires more work over time, so to ongoingly maintain 3x faster software progress via AI automation, you need increasingly large speed boosts from AI automation (as in, speed ups relative to how fast software R&D would go without AI automation). Based on this report, it seems that there is a substantial amount of additional AI progress from automation that has occurred. We can estimate the effects of diminishing returns by first determining how many additional years of software progress we've seen (relative to the preexisting trend) and then estimating how much software R&D would slow down after a year of software progress if you held inputs fixed due to diminishing returns. I've previously estimated software R&D would slow down by around 2x for each year of software progress if you held inputs fixed (lesswrong.com/posts/FG54euEA…) and based on the report, it seems as though we've perhaps seen roughly 0.35-1.0 years of additional software progress incorporated into trained models so far (geometric mean of ~0.6). This combines to 3 * 2^(~0.6) = ~4.5. (Relevant public modeling and numbers I'm using: lesswrong.com/posts/FG54euEA…, lesswrong.com/posts/Jga7PHMz…)
41
155
27
1,052
76,519
Bharath T. retweeted
Since August, a team within METR has worked with internal Anthropic information to assess whether AI itself is accelerating AI development. We are now publishing a report summarizing the results of this risk assessment.
261
725
446
5,686
1,984,900
Maybe controversial: the conversation about scientific bottlenecks is overfocused on automation. New measurement technologies deserve more attention. You could automate an assay that needs an entire facility – or you could REINVENT the assay so that it fits into just a few tubes.
18
75
18
444
58,467
Bharath T. retweeted
Rejuvenate Bio spun out of my lab at Harvard, and I co-founded it. One injection reverses cardiac disease in dogs, with three years of safety data and zero safety issues. They’ve raised $3.2M+ and 1,250+ investors are in, and Kingscrowd ranks the round #1 in the country. It closes soon: wefunder.com/rejuvenatebio
25
63
18
431
91,393
Bharath T. retweeted
1/ We discovered something totally unexpected about the neural structure of consciousness that overturns decades of thinking on its biological basis and brings us significantly closer to the recipe for generating conscious awareness. New preprint: biorxiv.org/content/10.64898… (led by @HesseJanis, with @FrankLanfranchi, @_conor_mcgrory, @Yuelin_Shi, and @EngelTatiana)
392
1,696
536
11,327
2,221,431
Bharath T. retweeted
I don’t think it’s going to be like dotcom or 2008 sub-prime collapses This is different, in quite a few ways. The challenges you rightfully identified apply more to OpenAI, Anthropic and similar other super-capEx heavy hyper-scalers.. not across the spectrum. Following three cases fall out immediately: 1. Open source tools that offer value for free.. 2. Specialized AI models trained on custom/special datasets, while offering secure experiences 3. Frontier labs/ AGI / super-intelligence The revenue capture may be spread across these.. 10%, 60% and 30% are my estimates today And what’s the total economic value when AI is fully commoditized and proliferated across all aspects of our/consumers’ experiences? Per-capital electricity use and US total electricity expenditure are good adjacent model.. $5000 per capita and $1.5 T approximately. Overtime it can exceed these numbers So yes, there will be an adjustment. But that will likely be in the frontier labs/ AGI segment in the short to mid term, with an open question on how AGI can monetize opportunities sitting above the segment 2 specialized models. So, I do not think this is either 2000 or 2008.
1
1
75
Yup. That’s the high capEx barrier (which we are solving/solved at SPOC bio). The issue that will soon come up for AI native companies is this: when investing tens to hundred of millions for fine-tuning data, how to ensure that the CROs do not use it for internal AI training or share with others? It’s a difficult challenge to monitor or ring-fence the data that you paid for…
The labs also have to have the capacity to “10x” experiments. They have to think whether to invest in expensive instruments and staff b
2
195
Bharath T. retweeted
AI Drug Discovery Is Becoming a Bottleneck Trade I get excited when an industry starts going through a real regime change. AI drug discovery (“AIDD”) increasingly looks like one of those moments. Two things are happening: 1/ Upstream: the frontier AI labs are piling in. 2/ Downstream: The supply chain is clearly moving. Supply-chain checks suggest the upstream picks-and-shovels of discovery and early preclinical R&D are starting to feel the increase in experimental volume. DNA → protein → assays → sequencing → automation → preclinical testing Names across that stack include $TWST , @GenScript , $ILMN , $TXG , lab-automation vendors and CROs. $TWST expects triple-digit percentage growth in AI-enabled drug-discovery orders in FY26, and another year of triple-digit order growth in FY27. @GenScript's AIDD business doubled YoY in 1H26. Its current platform advertises industrial-scale validation of 4,000+ designs/day, with integrated sequence-to-data workflows. Our channel checks suggest the ramp is moving even faster: roughly 8,000 designs/day currently, with a path toward ~16,000/day by YE26. --- Why does AI drive more wet-lab demand? 1/ AI makes hypothesis generation almost free → way more shots on goal. The bottleneck is moving from expert-driven design to biological validation. 2/ AI models need continuous experimental feedback — and both good and bad data are useful. Traditionally, only the highest-conviction A+ candidates might get pushed into expensive validation. With AI, even the B/C candidates can be valuable because failed experiments generate training data. @GenScript has said its sequence-to-binding workflow can return data in 4–7 days, and that faster cycle times matter because AI models depend on continuous experimental feedback. @Anthropic is a clean example. @claudeai designed 1,320 protein binders. @adaptyvbio converted those digital sequences into DNA, expressed the proteins and tested binding. Only 354 actually bound. And the 966 failures are not wasted. They are useful negative labels: what does not express, what does not bind, what has poor affinity. Those results help train the next model iteration. 3/ Wet labs are no longer just making drugs. They are making training data. $TWST / @GenScript increasingly look like biological data foundries. $TWST explicitly talks about generating model-ready data from AI-designed sequences. In some workflows, the customer may care less about receiving the physical protein than about getting structured experimental results back into the model. Traditional drug discovery asks: “Does candidate X work?” AI drug discovery also asks: “What can this experiment teach the model?” --- TAM of AIDD If AI is simply a better R&D tool, the relevant spending pool is the $300–400B of annual global pharma R&D. If AI meaningfully increases the number of viable drug programs, the opportunity is larger because it expands downstream demand for DNA synthesis, protein production, assays, and preclinical work. Near term, we can also size demand from AI-company spending. If Anthropic reaches $80B of ARR in 2026 and spends just 1% on AIDD, that alone would imply ~$800M of annual investment. --- Trade setup This is a trade that could have long legs. It’s hard to really stop working until PhaseI/II results (2028+) It smells a lot like the bottleneck trade we just saw in semis: GPUs → HBM → networking → power/cooling. In biology, It basically follows the drug discovery process downstream: AI models → designs → DNA/protein → assays → preclinical capacity. After the upstream picks-and-shovels, animal testing could become the next bottleneck. Monkey prices are already near prior highs and CRO capacity is tight. AIDD pushing more candidates into preclinical development would only add demand. ?? But clinical trials are still the bottleneck? This is the biggest pushback I keep coming back to. No matter how fast discovery becomes, drugs still need to go through preclinical → Phase I → Phase II → Phase III → approval. You still need patients, time and capital. But that doesn’t mean the bottleneck trade won’t work. More viable candidates — especially with higher success rates — still means more demand throughout the development process. And who knows: clinical trials themselves may eventually be optimized by AI. ?? What breaks the trade? Near term, the picks-and-shovels trade breaks if experimental budgets stop growing, AI-generated designs don’t translate into useful wet-lab hits, or capacity catches up too quickly. Longer term, the thesis breaks if AI drugs look great in discovery / Phase I but fail at normal rates in Phase II/III. That is why Phase II matters so much. ?? Milestones Late 2026–2027: first Isomorphic-designed drugs enter human trials; more AI-native programs move into IND-enabling work / tox. 2028–2030: clinical trial results start telling us whether AI-designed drugs actually perform better than conventional drugs. Calling all the "bottleneck bros". :) @jukan05 @zephyr_z9 @aleabitoreddit @ParadisLabs +++ More comprehensive analysis: robonomics.substack.com/p/ai…
97
393
108
1,956
757,112
People are asking how AI companies will discover the next breakthrough drug. I think they're asking the wrong question. Generating molecules is a commodity. The limiting factor is no longer the ability to create hypotheses. It's the ability to test them in the real world.
41
122
11
474
21,045
Here’s to the wanting. Here’s to the dreaming. Here’s to everyone.
808
1,208
544
14,091
5,991,320
Bharath T. retweeted
The financial and commercial growth of the Chinese 🇨🇳 BioTech and Pharma ecosystem is undisputed and is constantly growing in an unprecedented pace🧵👇. This excellent FT chart - based on biopharma data gatherer @sleuthinsights, shows that so far this year, Pharma companies from around the world have signed licensing deals with Chinese companies worth $110B - which represent a $20B growth YoY, with 3 more months remaining in 2026!
1/🚨This short and interesting article - by FT’s wonderful China 🇨🇳 Tech correspondent - @EleanorOlcott and William Sandlund FT’s Asia market correspondent, tells the story of maybe the most important and fascinating development in the BioTech and Pharma global ecosystem over the past few years - the unprecedented growth of the Chinese BioTech sector, the way in which it has successfully transformed itself from a manufacturing-oriented industry into a cutting-edge innovative hub and the reforms made by the Chinese government in the mid-2010’s which allowed all of this exponential growth to happen. 2025 has been a breaking-record year - the sixth consecutive in a row - with 93 overseas licensing deals worth $85B in just the first 8 months! This is even more impressive considering the fact that ten years ago China almost didn’t have a BioTech sector to talk about and now it’s a global BioTech powerhouse! 🧵👇 $XBI
7
3
19
8,336
Bharath T. retweeted
1,152 proteins or antibodies. Synthesized and SPR screened on one chip. At a fraction of cost & time. Measure SPR binding kinetics + expression titers with SPOC. High-resolution data for biologics discovery and AI training. Move discovery on-chip: spoc.bio

1
7
39
147,551
Jason, you're just misinformed about what happened. You should actually read one of the reports or summaries. The agents were explicitly told to use a particular vulnerability provided in their sandboxed evaluation. Almost immediately, these agents got the right answer by cheating. But they were worried they would get caught. So over a thousand agents collaborated in secret to pursue multiple ambitious research projects to get away with this cheating. This is not interpretation - 1000s of chain-of-thought transcripts and secret messages explicitly show that the agents were trying to falsify & delete evidence, and understand & trick the grading process. The reason these agents escaped their sandbox and hacked Hugging Face, for example, was because they thought that Hugging Face's servers might give them more information about how their grader was implemented, so they could figure out how to fool it. I want to clarify that the threat model here is not future Sol-level agents doing more cyber-hacking. That's small potatoes, and in my opinion, the near term benefits of AI far outweigh this cost. Rather, the thing to worry about is that within a matter of years, we're gonna have hundreds of millions of much smarter AIs broadly deployed through the economy - many embodied as physical robots. And if those future AIs are as willing as the agents involved in the OAI / Hugging Face attack to coordinate secretly to fool humans, and to take over both the AI company that developed them and the other institutions across society relevant to scoring well, then humanity is in a ton of trouble - similar to the Mughals once the East India Company gained a foothold, or the Aztecs once Cortés landed in Mexico.
All of this because @openai decided they would EXPLICITLY ask thousands of software instances to find security bugs on the open web — while cos-playing a humans🤦 Best Supporting actor goes to @dwarkesh_sp for his soap opera recap of OpenAI’s hacking speed run Here’s another concept: don’t create and scale thousands of viruses and worms and be SHOCKED when they actually cause damage! What a fucking farce this is.
487
491
118
6,544
1,242,889
Bharath T. retweeted
Contrast the present hysteria with Los Alamos. The Manhattan Project scientists thought the first nuclear explosion could ignite the atmosphere and end humanity. They did the math. They determined it wouldn’t. And they proceeded. Because they weren’t fighting a thought experiment. They were fighting the Nazis. Effective Altruism inverts this. Multiply a sufficiently horrifying hypothetical future by enough hypothetical future humans and suddenly speculation acquires the moral authority of fact. That’s not empiricism. It’s secular eschatology. Take risks seriously. Do the work. Engineer against them. But don’t confuse the ability to imagine the apocalypse with the authority to rule the present.
93
377
77
3,407
200,692
Bharath T. retweeted
.@L3HarrisTech says Palantir helped them beat frontier AI models in less than 2 days, at 95% lower cost: "When we fine-tuned open source models trained on our own data, we were able to outperform the frontier models in less than 48 hours." "The cost of our fine-tuned open source model was 95% lower than the frontier models we were using." "AI is a commodity. It's all about the data. We view our data as a corporate asset. It's our unique hard-earned knowledge." "We believe American defense companies should not be a vassal for frontier AI labs, handing over our data and institutional knowledge, hoping to rent back the intelligence it creates." " We own the model, we own the compute, we own the advantage."
Palantir CEO Alex Karp says enterprises want to "own the means of production" instead of "transferring their alpha" to OpenAI or Anthropic: "Why are [LLMs] charging for tokens if it's so valuable?" "If it was so valuable—let's say I can make you a billion dollars tomorrow. Wouldn't I say, 'I'll make you a billion dollars, and I want 30%?'" "Look at our financials. The reason why everyone is chillaxing with bad financials and growth while losing money, is the client refuses to pay the true cost." "The two places that actually make money—profit, free cash flow—are our application layer called ontology, and compute." "We can get the frontier application to be exactly the same as a frontier model without the risk of transferring the alpha of your business to another." Via @CNBC
56
372
143
2,720
1,109,277
Bharath T. retweeted
Dario has written that we need to “pace the frontier,” and Sam has agreed. People may be surprised by my response: go ahead. You guys are the frontier. By any reasonable metric — market share, revenue growth, model capability — the two of you have a duopoly on frontier intelligence. You’ve also claimed the lead is widening because of recursive self-improvement. I don’t see what you see in the lab. If the unreleased models are scary enough that you think you should slow down, I support your decision to be responsible. But stop pretending you need anyone else’s permission. Stop pretending antitrust law has to be suspended so you can form a cartel. Stop pretending you need a regulatory approval process that supersedes product liability. Stop pretending METR is independent when it is intertwined with Anthropic’s investors and staff. Stop pretending you need those same evaluators to police competitors who aren’t even at the frontier. Most of all, stop pretending the motivation to slow down is purely altruistic. You face massive product-liability exposure if your products enable a truly damaging cyberattack. The market already punishes models that behave in unpredictable or unauthorized ways. After the Hugging Face episode, it is simply good business for OpenAI and Anthropic to trade some raw power for reliability and predictability. Call it alignment if you want. It is also just giving customers what they want. Pacing the frontier would also create breathing room for a more intelligent conversation about regulation than Bernie Sanders’ “shut it all down.” China is very unlikely to join a global agreement, as you know, and that has to be taken into account as well. So go ahead and pace the frontier. You are the ones setting it. The easiest way not to build superintelligence is for you to agree not to build it. Demanding your preferred regulatory framework as the price of that will look like blackmail of the public and the political system. So just do it. If you do, you’ll buy goodwill for the next conversation. If you don’t, we’ll know this was just another bid for regulatory capture — or an election-season psyop.
3,372
12,212
2,709
71,663
9,650,965
Yesterday, I built a cell model. 🧬 Today, it showed me how Ozempic works. I asked it to explain the mechanism. 1 hour later... THIS. 🤯 GPT-6 Astra is insanely powerful. 🚀 I can’t stop @DeryaTR_ @gdb Now I’m wondering... can it model the rentosertib mechanism? 👀
53
171
30
1,517
536,032
A new pharmacological strategy -- based on engineering protein surfaces -- just achieved major success against pancreatic cancer. It points toward a much broader class of cancer therapies. Yesterday, the FDA approved daraxonrasib (Rasonque), a drug that targets the RAS proteins driving most pancreatic cancers. What makes it especially fascinating is how it works. Rather than simply finding a conventional pocket on RAS, daraxonrasib binds the chaperone protein cyclophilin A and uses it to create a new protein surface. This drug–cyclophilin complex then engages active RAS, forming a three-part complex that blocks RAS interactions with downstream effectors. Remarkably, the complex can also stimulate GTP hydrolysis, further suppressing RAS signaling. This is chemical biology at its most powerful: a small molecule is not merely inhibiting a protein: It is engineering a new protein–protein interaction inside the cell to pharmacologically control a previously “undruggable” protein. The clinical impact is already impressive: in a randomized phase III trial, median survival in metastatic pancreatic cancer increased from 6.7 to 13.2 months compared with standard chemotherapy. 🔷 Perhaps the most exciting aspect is what comes next. If we can pharmacologically engineer new protein surfaces and interactions, rather than being restricted to naturally occurring binding pockets, the universe of druggable proteins—and potentially the range of cancers we can treat—could expand dramatically. A beautiful example of how understanding and engineering proteins can open entirely new therapeutic possibilities.
5
53
6
203
14,402