Brae

Chemist, engineer, programmer, PhD. (he/him)

[Brae @braev_pete]

@matthewhirschey I've wondered if a slack/text channel may be better arena for exchanges about a paper. Starts with a comment question that people can delve deeper into as they wiki-spiral more into that thought. Supports a slower advancing, but perhaps deeper discussion.

[Brae @braev_pete]

@mealreplacer https://t.co/tyzkGLwm6m (previously https://t.co/PoKwaco6zJ) has been okay at that so far.

[Brae @braev_pete]

@neuralreckoning @overlordayn @pfau When you say it out loud, it really baffles the mind how the scientific publishing system works. Your right, I haven't met any scientists who like paywalls... Yet here we are.... Haha (😅🤔😭).

[Demetri (is over at the other place too) @PhDemetri]

Sample size calculations become a lot easier when you realize "Hey fuck it, I can just simulate this"

[Brae @braev_pete]

@yoavgo @marian_nmt Yeah, I can see the bogus it comes up with at some times, but also, it pulls an idea that was mentioned in passing in a book or a class 10 years ago and i go, ooh that's right.

All about using it the right way for sure.

[Brae @braev_pete]

@levikul09 Languages, ai, data handling will continue to develop and change: problems will always be around! Knowing how to solve a problem, you can see how a new/old tool can help with it.

[Brae @braev_pete]

@LabPresse Lol that's cool. I think we are even better learners when it's around a familiar context (especially something abstract like improving writing flow/conciseness). For one (me) whose writing could improve, I'm excited to try that out to see if I improve generally. Neat.

[Brae @braev_pete]

@marktenenholtz From another perspective, a model that is 20% wrong is still 80% right, that may be good enough!

I think GPT API can get 80% right in a lot areas, which is good enough for a lot of problems!

[Brae @braev_pete]

@marktenenholtz 100% Understanding the needs can get glossed over when there is an excitement about new things. It could be your needs dictate a model a lot simpler.

A model's usefulness isn't determined by how complicated or how novel it is. But does it meet the needs required.

[Brae @braev_pete]

@MilesCranmer It is a wonderful tool for brainstorming and working on flow, and now you can do it when your pulling a presentation together at 6am for a 8am meeting ;)

[Brae @braev_pete]

@sama Yes! I've used it a few times to reword ideas from a journal article that was a little too obtuse. Very handy for improving understanding on either end.

[Brae @braev_pete]

@michael_nielsen It concisely spits out the big ideas, handing me the equations that describe the major factors that will influence my system of interest I can interrogate more thoroughly *in silico*.

[Brae @braev_pete]

@michael_nielsen Give me the top 10 Mathematical models to describe [insert current research problem]. List the models in order of most to least popular.

[Brae @braev_pete]

@jimhaseloff From a quote below-- I am excited for scientific oriented or specialized consumer LLM like a "sci-chatGPT". Let the PhD candidates quickly rubber duck ideas, and iterate them with AI. A more efficient way of building off previous work. Could be part of the change hoped for here.

[Matthew Hirschey💡 @matthewhirschey]

100% of research labs have LAB MEETINGS.

0% are set up to enable everyone to have a voice.

This is how we flipped them to
increase engagement
hear from everyone
and save time...

[Brae @braev_pete]

@ATinyGreenCell 😂 I asked it for insulin earlier today and got a an extensive loop of Lys and Ser.

Crazy that with the right training/different model you can get text to proteins tho!

[Brae @braev_pete]

@StearnsLab Encourage problem solving and discovery. Ability to synthesize knowledge so easily will be amazing for that.

[Brae @braev_pete]

@sama More time to play with it :) it has been really interesting using it as a tool for some chemistry/bio simulation starting points. Speeds up the beginning research and prototyping phases for sure.

[Brae @braev_pete]

After playing with chatGPT this morning I think it is going to be as useful to scientists as Google over the next decade. Being help explain an article or bounce ideas off of. Part of me hope I'm wrong, but new tools and new ways to improve my workflows.

[Brae @braev_pete]

@lanky_professor @For3JT BYUs was always too busy when I was there lol.