←
Steven ten Holder1,548 posts
Steven ten Holder avatar

Steven ten Holder

@steventen

building zeroshot bio // previously co-founded acorn biolabs, YC 2016

◷ Joined September 2012▧ 116 Photos and videos
Steven ten Holder avatar
Steven ten Holder@steventen·Nov 12

Progress in ai x bio will depend on contributions from many talented teams exploring a variety of slices of the bio stack. @hsu_jonny and his team @blankbio_ are pushing the RNA frontier:

Today, we are excited to introduce Ladon, a state-of-the-art biological foundation model that predicts disease trajectory and treatment response in patients. There is immense complexity in the individual biology of every patient. Two patients can have the same diagnosis and receive the same treatment, yet have very different outcomes: one responds well and lives for years, while the other’s disease continues to progress. Understanding and resolving this heterogeneity to predict outcomes is how we accelerate our delivery of medicines for patients and unlock clinical trial abundance. Ladon is our first step towards this future: one where we can learn more from every patient enrolled so that we can run smaller, faster trials with higher success rates. We built it to give drug developers quantitative support on billion-dollar clinical trial decisions. Today we're excited to share Ladon's capabilities in our blog and technical report. Ladon is a 3B parameter model trained on millions of RNA-seq samples spanning decades of technical and biological variation across tissues and species. Ladon beats gene expression baselines, established biomarkers, and existing RNA foundation models at predicting patient prognosis and therapeutic response in 140+ oncology tasks, including external validation on past clinical trials. This is significant because clinical trials are fundamentally generalization problems. We show that Ladon-derived predictions are robust across indications, treatments, and sequencing platforms. Reach out if you’d like to get started with the model today! Read more about Ladon in the blog: blank.bio/introducing-ladon Find more technical details in the white paper: blank.bio/introducing-ladon

Post attachment
Steven ten Holder avatar
Steven ten Holder@steventen·Nov 8

All the encouragement one needs to work in AI:

AI researchers seek to understand intelligence well enough to create beings of greater intelligence than current humans. Reaching this profound intellectual milestone will enrich our economies and challenge our societal institutions. It will be unprecedented and transformational, but also a continuation of trends that are thousands of years old. People have always created tools and been changed by them; this is what humans do. The next big step is to understand ourselves. This is a quest grand and glorious, and quintessentially human.

Steven ten Holder avatar
Steven ten Holder@steventen·Nov 7

We‘ll never ‘solve’ biology for the same reason quantum uncertainty retired Laplace’s demon. What we will do is improve our theories-of-biology. Our models will increasingly reflect the patterns of life, hopefully toward useful new bioengineering and drug discovery tools.

Each of the 400k particles is a unique, untrained, randomly mutated ~10k param neural net btw 🤗 🔊🦠 Neural-Kimchi-4B 🔊🦠

3:48
Steven ten Holder avatar
Steven ten Holder@steventen·Oct 21

“An approach we call distributed biological reasoning.” Slowly but surely, AI is absorbing the shared intuition of the world’s great biologists. We won’t *just* get valuable new findings from existing bio data — we’ll also begin to more deeply understand how to make highly effective *new* bio data. So exciting!

OpenAI@OpenAIOct 21

GPT-Rosalind surfaced a new lead for drug discovery: could restoring cholesterol transport protect neurons from degeneration? DCTN1 is a gene linked to Perry syndrome, a rare neurodegenerative disorder. In mouse-brain data, disrupting it increased expression of genes involved in making cholesterol. GPT-Rosalind suggested a possible explanation: cholesterol may not be reaching where neurons need it. That gives scientists a specific mechanism to test and, if confirmed, a direction for developing new treatments for neurodegenerative disorders. With @perturbai_tx and @scrippsresearch, a team of thousands of GPT-Rosalind agents worked in parallel, connecting gene-expression results from millions of mouse brain cells with scientific literature in just 32 minutes. We’re now expanding this collaboration to identify new drug targets. openai.com/index/scaling-ai-…

Steven ten Holder avatar
Steven ten Holder@steventen·Jun 14

Our team just minted a 1-million-cell drug-response dataset.. in zebrafish. I'll be in San Francisco Monday, Tuesday and Wednesday this week. I'd love to meet AI folks interested in modeling biology. DMs open

0:10
Steven ten Holder avatar
Steven ten Holder@steventen·Jun 10

Well articulated 🫡 Building toward a true 'bio-generalization' North Star is what gets people excited. Recalibrating expectations toward mini-generalization wins along the way is the road to effective progress. Ideally, we can balance hype with bite-sized utility year-by-year

We need to continue to be honest about what our models are actually capable of and where they fail so we keep moving forward. Less hype & blind faith in parameters & arbitrary scaling. More evidence based strategies & honest reflection. 6/6

Steven ten Holder avatar
Steven ten Holder@steventen·Jun 4

With all of the money and research pouring into AI, I'm surprised we don't have more awesome (in the original 'eliciting wonder' sense of the word) visualizations of what trillion-parameter neural networks really 'look' like on the inside..

This is what the AI brain looks like.

0:21
Steven ten Holder avatar
Steven ten Holder@steventen·May 14

Yeah! As I've begun to develop an intuition on the limits of AI, I've had to incorporate concepts as far out as the philosophy of knowledge, as 'practical' as brain surgery, and as basic as statistics.

Hot take: machine learning and AI did more to understand the nature of knowledge, and our relation to reality than 20 centuries of philosophy. I am ready to kind of defend this hill.

Steven ten Holder avatar
Steven ten Holder@steventen·Apr 27

I bet we’re going to discover many very useful abstractions on the long road to ‘full AI’ understanding. Some of biology’s most reliable, important patterns won’t require atom-by-atom reconstruction to be effectively learned.

This tweet is unavailable

Steven ten Holder avatar
Steven ten Holder@steventen·Apr 12

If you enjoyed reading Isaac Asimov's robot stories, I guarantee section 7 of the Claude Mythos Model Card will tantalize you. It reads like Dr. Susan Calvin -- Asimov's fictional 'chief robopsychologist' -- dissecting the mind of yet another robot gone awry.

Post attachment
Steven ten Holder avatar
Steven ten Holder@steventen·Mar 24

couldn't agree more. 1. human relationships 2. artistic creation 3. scientific exploration these three will remain the most engaging, meaningful categories of human activity post-ASI (i think..)

I think most people should consider becoming scientists. While there is a real risk AI will eliminate jobs, conversely the need for models to understand the world around us through more and better data will increase exponentially. AlphaFold, for example, would not have been possible if not for the work of thousands of grad students and postdocs painstakingly depositing well-curated datasets over decades of meticulous experimental work. We still only understand a fraction of the physical world and beyond, and we’ll only be able to do so if we collect more data and put it into context. Maybe this is a bit naive or utopian thinking, but humans are not designed to look through pages of receipts, read 100-page legal documents, or write reports and POCs no one will ever read. They are wired to explore, driven by curiosity, and have an innate desire to understand the world. Many who are bored or feel a lack of purpose -- with the overhanging (and maybe exaggerated) doom and gloom of AI replacement -- should think about this.

Post attachment
Steven ten Holder avatar
Steven ten Holder@steventen·Jan 27

How long before lab technicians are prompted through protocols by lab-AIs? Scientist-in-the-loop.

claude is independently connected to a plate reader and writing a PyLabRobot interface to it

Post attachment
0:19
Steven ten Holder avatar
Steven ten Holder@steventen·Jan 9

We need even more enthusiasm-fueling art. Using great art to communicate the 'awe and wonder' behind why we build not only helps society-at-large understand our motives, it also makes the journey that much more aesthetically wonderful for ourselves. Thank you A16Z 🫡

a16z@a16zJan 9

It's time to build.

1:46
Steven ten Holder avatar
Steven ten Holder@steventen·19 Dec 2025

More intelligent models should not only be capable of producing simulations like these, they should also be capable of evolving the environments and variables themselves toward undiscovered categories of useful new patterns. "artificial interestingness selection"

AA@measure_plan18 Dec 2025

the secret life of boids using gemini 3 flash to simulate simple rules that lead to emergent behaviours (attraction, repulsion, cohesion)

0:30
Steven ten Holder avatar
Steven ten Holder@steventen·11 Oct 2025

Science fiction is science vision. The very first short story in I, Robot is about a young girl’s heartache after being separated from her loving robot.. Surreal to see it happen for real this year:

Post attachment

A Chinese father's video of his daughter tearfully saying goodbye to her broken Al learning robot. People already get emotionally attached to AI. Now imagine Figure03 at home and it breaks, people will have a breakdown.

1:07