Lukas Hafner

Biologist lost somewhere between the scales | Biology x ML
@TechnionLive
previously @BIUPasteur @lpiparis_ @Biologie_UNIGE

Opinions are my own& mostly wrong

[Lukas Hafner @LostInTranscrip]

Excited to see our data-to-paper framework out @NEJM_AI !

https://t.co/LEMuAFzP82

[Roy Kishony @RoyKishony]

Our data-to-paper AI is officially out!
https://t.co/wpKcdAElKr

Autonomously analyzes data and writes human-verifiable papers!
Fun AI-human copiloting!

Papers are "data-chained": click-traced to the raw code.

Try it out:
https://t.co/yhqDfTj4aB

@TalIfargan @LostInTranscrip

[Lukas Hafner @LostInTranscrip]

@erika_alden_d Maybe you should send them a deck of ideas for a new extension package!

But I guess it is cooler anyway to "actually" terraform mars 🚀

[Lukas Hafner @LostInTranscrip]

Feels like a #terraformingmars card

[Erika Alden DeBenedictis @erika_alden_d]

Today, Pioneer Labs is announcing our first step towards terraforming Mars. 🚀🌼

With equipment that fits in just a single rocket launch, we can convert Martian dirt, water, and air into enough building materials to construct a small city on Mars.

To do it, we made the first microbe for Mars. We found the best microbe on Earth and used evolution to teach it how to source all of its nutrients directly from Martian materials.

The first astronauts will be greeted with safe shelter already filled with water, oxygen, and rocket fuel for the return journey.

This is the first step toward using biology to make Mars a friendly place for life. It lets us live off the land and helps us build the next great frontier.

It's the first of five organisms we need to green Mars ⬇️

[François Fleuret @francoisfleuret]

We speak at ≈ 8k words per hour.

300b tokens ≈ 230b words ≈ 10'000 years speaking 8h per day.

[Lukas Hafner @LostInTranscrip]

@roxannevarza It is the pay-off from fundamentally thinking over their education system ~25 years ago

[Lukas Hafner @LostInTranscrip]

@nntaleb The baker lab? Which one exactly?
@UWproteindesign ?

[(((ل()(ل() 'yoav))))👾 @yoavgo]

re "disclosing AI use in papers" discussion that re-emerges: please lets not do it

as a reviewer or a reader of a paper, i do not care what part of the ideation, research, proof, implementation or writing was done by AI, and how the credit is split between the AI and human author.

at the end of the day, humans sign on the resulting paper, and this is all i care about. that they stand behind the text they produced. i don't care if the idea is theirs or came from the AI or is a shared process. i do care that they found it compelling enough to publish.

in any case i don't trust the authors to accurately describe how the idea evolved between them and the AI: even if they do try and be honest, i think in most cases its a complex process the author themselves don't fully understand.

when i work with a student, i certainly do want them to explain how they used AI and which parts of the work, idea or argument are primarily AI driven. but this is in order to guide and improve the student's work and process, not because i care about the credit assignment. it's part of their training and my mentoring, and should not leak into the final work. the readers and reviewers should not be bothered with it.

so please. let just have people submit work without the silly AI disclosures, and be very trigger-happy on rejecting papers that are clear AI slop, or that are hard to evaluate due to what may be improper AI use.

(in the short term, we will likely see an increase in conference submissions that have the same idea, coming from different authors having a similar process with a shared AI. i think this will sort itself in the long run by reducing the credit assigned to such works in the community, but in the short term we may want the conference management software to actively look for such similar submission, and maybe review them in bulk)

[Roy Kishony @RoyKishony]

This is so inginius (and I am worried likely true): "in the future our entire work will just be to prove we are human". What a cool and scary idea; perfect for a Black Mirror episode!

[Lukas Hafner @LostInTranscrip]

5/ Title: CAPTCHA.

Story follows one person's offboarding — and what they rediscover about being human once they're finally outside the system.

I have an ending/twist in mind but won't spoil it here (yet)👀

[Lukas Hafner @LostInTranscrip]

@netflix
1/ Black Mirror episode pitch: In the future, AI does all the work. Humans just live off universal basic income.

Except there's one taks left: proving you're not a robot.

And it's slowly eating people's entire lives (if you have a phone, maybe you can relate!)

[Lukas Hafner @LostInTranscrip]

4/ Here's the twist: the offboarded aren't just surviving in the shadows of the cities. Freed from the relentless grind of authentication, many of them are thriving! Poorer, but freer - while the "IDs" spend their days trapped in self-driving taxis, buried in verification loops.

[Oded Rechavi @OdedRechavi]

Most experiments fail, and negative results rarely get published. This means LLMs are unaware of the outcomes of most experiments.

[Lukas Hafner @LostInTranscrip]

@caldarellip1 Interesting. But I think the capacity for direction and responsibility will be just completely dependent on the ability of being able to judge in the first place.

Also, to what degree can taste be learned without the need to execute (as it is so cheap)?

[Lukas Hafner @LostInTranscrip]

With AI getting better and better: will "taste" become the "key" human skill; e.g. the sense of knowing what is "good" and/or "true"?

[Umberto Aiello @UmbertoAiello]

1/ Do you have a favorite protein you wish you could dissect residue by residue? 🔬 Excited to share our platform for mutational scanning at endogenous loci in yeast (no ectopic expression needed!) https://t.co/hCs3TGrfay

[Rayan Chikhi @RayanChikhi]

🌎👩‍🔬 For 15+ years biology has accumulated petabytes (million gigabytes) of🧬DNA sequencing data🧬 from the far reaches of our planet.🦠🍄🌵

Logan now democratizes efficient access to the world’s most comprehensive genetics dataset. Free and open.

https://t.co/dDBtAjfdYL

[Yunha Hwang @Micro_Yunha]

How do we build AI systems that enable deeper, not just faster, science? I came across a very thought-provoking article by @nisheethvinoi on “What Counts as Discovery?”

[Michael Baym @baym]

Legitmately thrilled to share our latest work, in which @fernpizza solved an experimental challenge in plasmid biology as old as the field: measuring how plasmids compete and evolve within individual cells!

[Paolo Caldarelli @caldarellip1]

📢 First paper from my Postdoc is out. The field of stem cell-based embryo models is flourishing. These models mimic critical stages of embryo development, providing powerful tools to study processes that are tricky to dissect in natural embryos. Many approaches are being used to investigate them—what was ours? 👇
https://t.co/MlJ8PUzlSl