Dan Ofer (✈️ ICML-26 Was @Worldcon)

1/ Our paper “Protein Language Models Expose Viral Immune Mimicry” is now published in Viruses! We show that protein Language Models can identify viral proteins, and those that fool our immune system.

@random_walker I enjoyed it.

I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: cs.princeton.edu/~arvindn/talks… I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”.

@houchao1 I'd argue they mostly don't, except for structure. We did that in puberty, I was hoping to be disproved by now 😭

🚨 Our paper of explaining why larger pLMs do NOT always perform better on fitness/variant-effect prediction was published. See our previous thread for the key insights, and check out the paper for the full story.

@jueseph What about function and novel sequence understanding?

The protein design team at Deepmind is hiring! We are looking for someone to run design campaigns and contribute to model improvement. Should have expertise in design applications and using wetlab data feedback to improve methods. Please RT! DM me with Qs. google.com/about/careers/…

@pkoo562 You forget to not compare to any non deep baselines

Anatomy of BioML papers: 1. create benchmark misaligned w/ REAL biological goals 2. develop ML model w/ jargon-maxxing 3. compare vs models that are inappropriate or out of context 4. give it hypiest name — foundation model 3 years ago; virtual cell last year; world model now

@airkatakana Stan a better automl

tpot finally reuniting after 3 years

Any questions? :)

1/ Our paper “Protein Language Models Expose Viral Immune Mimicry” is now published in Viruses! We show that protein Language Models can identify viral proteins, and those that fool our immune system.

@Aella_Girl Nimby seems to sum it up

Why and how is it like this? Like clearly something has gone wrong in the sf government but what went wrong specifically?

@cloneofsimo I'm still around. (you rando) 😅

Guysssssss stay in seoul for longer 🥲🥲🥲🥲

@celestepoasts Japan teamlabs

where I post from

@airkatakana Also on comiccon style conferences. Very common.

how chinese do you have to be to take your parents with you on your business trips

@airkatakana Socialising and showing your research stuff that isn't faang lab adjacent

i would have loved going to ml conferences in 2015 when i didn’t know anything about research and humanity wasn’t sure how to build agi but now i know 99% of academic research is fake and most of the real stuff is behind closed doors anyway what is there to get excited about

That's a wrap on this year's ICML. Fun and stimulating! To all those I met - happy to keep on talking science (and the rest). :)

@algobaker @RobertTalisse I'm in it for the science. Socialising is just an unplanned, optional bonus

@danofer @RobertTalisse Might not be the best choice of conference if you want to try academic conference hookup culture haha

I could have lived a generally happy and fulfilled life without ever having heard about "academic conference hookup culture"

Presenting our protein LLM/ML papers today at ICML #3400 Drop by 📸 [ICML] Int'l Conference on Machine Learning

@Aella_Girl Ds yes, pro no. (lacks employment or consulting history)

Do I (aella) count as a professional data scientist?

@dejavucoder The wait list was impossible for that

if i knew icml was gonna have kpop performances an year back, i would have published a paper and gotten my ass sponsored by someone to attend

@Sep_Rahimian @RobertTalisse Ones with people under the age of 70?

@RobertTalisse This is disgusting and unprofessional. What kind of conferences have this culture? 😁

@OdedRechavi I'll fix my headshot for next time

Academic seminar posters that include a headshot and abstract should really be designed like WANTED ads