Jase Gehring

Many such cases

We could have a space elevator before the decade is out, but chemists decided it wasn’t worth it to share their raw data + metadata in machine readable format

custom tailored lab coat. gigaflex. necktie always looks good under a lab coat. this man has fucked around and he is about to find out in the best way

you think LLMs have powerful world models? no. no man has handwriting this good, and piss-scribbling of this quality is simply out of the question. a truly intelligent system would know that

why has machine learning been so much more successful in biology than chemistry? because biologists, in the face of staggering complexity, made their data publicly available and machine readable. chemists, meanwhile, decided a PDF of an NMR trace was good enough. a costly mistake for humanity

Super important structure. Congrats and thanks

Gonna assume this is just a joke

very interesting. i've had similar success with misdirection in bio domain. i'm not making unsafe requests, just bypassing overly stringent safety filters. this legitimately unsafe request takes it to another level

any excuse to use the glass cutter! i cut them in thirds and THEN i need to etch the backside 🗡️

mmhmm ya i'm doing surface modifications and it's kinda nice. i also like to etch or notch corners for the same purpose. any excuse to use my glass cutter 🗡️

that's the transition i'm talking about. our industry is already heavily automated, and there are lots of places LLMs can fill in an reduce human intervention. i think it will be a surprisingly smooth progression as we are able to offload more of the scientific cycle to AI

Hey Nick! Hope all is well with you and the escalante crew!! 🚀

glass slide with G L O B E printed on top, "unmod" written on bottom in scientist scrawl

how do we feel about glass slides with a manufacturer's logo on?

big implications for open models. most users will be able to do a ton of useful work with just local inference

i often get lost in the drudgery of labwork when really i'm using so many skills that are still uniquely human. for full automation, we need systems that combine LLM, world model, real-time visual motor integration, and fine dexterity. i'm angling to start working on this transition

Still only ~5% of the NIH annual budget, but overall China is on an incredible scientific trajectory.

it is direct but also pretty dishonest and/or misleading!

Same. would be awesome to have direct integration with a vision/language model! someone probably has this going with local inference already

Ya or preview or something. I don’t think it’s available to me and I’m tired of navigating their mess to find out. very Google problems

I get Gemini from work and I’ve been waiting to use G3 in the CLI but it’s been two months and it’s still not available. Huge waste of time setting up a google cloud project

It can really code. Feel like they’re just behind on the harness

‘Sticks’ ‘By Element’ FTW

The world would be a better place if everyone used this representation of DNA at all times

If and only if!

I’ve transitioned to using LLMs as interfaces to other tools. LLM +codebase, +docs, +software, +web search. Massive increase in utility over a chatbot

The base LLM without context has many issues. An LLM with appropriate context and guidance is much more powerful

“Well, look, I’ll never say no, and we will always investigate any evidence of misconduct, BUT…”

- Todd Blanche, Deputy AG

Tech companies insisting “there is no bottom.” regulators urging “keep looking”

It’s better because when the work is done there is just more work but when the agent framework is done there is … ah, no, still more work

Wait I think this is a great candidate for natural language interface! CAD/Blender too!

Something I’ve often thought about but never really learned about. Will read with interest!