James Schloss (Leios)

Day at the zoo with my son

I don't know if it is possible to do an "on-topic" research stream. The very point of literature review is to bounce between a bunch of disparate concepts to create new ideas.

When you are working on an experiment, well, it's a bit dry unless you talk about a bunch of related concepts as well.

I like typos. We live in a society that has a constant need to auto-correct or polish their texts, but I don't even use spell-checker half the time.

It doesn't matter if it's an email or a private message. I want to hear your thoughts, not the thoughts some AI thought I would like

By the time vulkan shaders are done processing, I have lost any enthusiasm I might have had to play the game to begin with.

Stop. Just stop. There is no situation where I want an AI to write for me.

This is on my youtube community tab, by the way. If you are getting a bunch of quick reactions from your favorite youtubers, they might be using this feature.

Working on my animation engine in #julia

www.leioslabs.com/Fable.jl/dev/
www.youtube.com/c/leiosos

Everyone talks about AI like it's a world-ending threat, but it's mid at best.

Is `import numpy as np` also affected by tariffs?

GitHub - JuliaMolSim/Molly.jl: Molecular simulation in Julia Molecular simulation in Julia. Contribute to JuliaMolSim/Molly.jl development by creating an account on GitHub.

A big new release of Molly.jl just dropped.

Including improved performance, especially on GPU (from Giuseppe Gambini), and support for other GPU backends (from @leioslabs.bsky.social).

github.com/JuliaMolSim/...

White House Budget Office Orders Pause in All Federal Loans and Grants The full extent of the order was not immediately clear, but the directive sent to government agencies on Monday threatened to paralyze a vast swath of federal programs.

www.nytimes.com/2025/01/27/u...

Yup, this is it. My absolute worst fear realized.

Every technological development in history has been the result of research. It has been the backbone of the US economy for ~75 years.

This is the type of thing that will truly break the US.

A researcher's perspective on Trump YouTube video by Leios Labs

I did it. I talked about politics on stream. I do not intend to make a habit of it, but here is one researcher's rambly opinions about Trump: www.youtube.com/live/G9bXl-k...

Trump’s team looking at withholding massive research grants from ‘woke’ schools Trump’s nominee to head the National Institutes of Health, Dr. Jay Bhattacharya, wants to target so-called ‘cancel culture’ at top universities

The US has been "on top for so long" because we have been taking the best and brightest people from around the world to do research.

If we cut fundamental research, we are destroying the foundation of the US economy.

Context: www.independent.co.uk/news/world/a...

Every now and again, I get a comment from some random person on Reddit who found my old writingprompt responses and wants to read more.

I then spend hours just tip-tapping away at my keyboard, writing a novel for some random stranger that will likely never see the final product.

Need #GPU help on a #research project?

Let me know. If it's interesting enough, I might pick it up for LeiosLabs in 2025.

I'll be discussing whether it is actually worthwhile to pursue a #PhD tomorrow: www.youtube.com/live/nT6tsC1...

If you have specific questions, feel free to ask me here.

Your research is only as good as your ability to communicate that research.

#climate discussion from a (previous) climate researcher... kinda.

Live now:
www.youtube.com/live/d3dV2nm...

I gotta be honest, this app is really refreshing.

More analysis here: www.youtube.com/live/a6gj7E4...

Which is very, very far from the claim that "AI-generated poetry is indistinguishable from human-written poetry and is rated more favorably"

I know it's just another AI bro paper, but it sucks that almost all the papers I read are like this now.

I think a proper comparison could have been made if they compared the AI poetry to authors known for having good meter and rhyme like Poe or Suess even.

But as it stands now, all the paper is saying is that "people like meter and rhyme"

No duh people are going to like a poem with meter and rhyme to Geoffrey Chaucer, a poet from the 1400s when middle english was all the rage.

I mean, what even is this list?

Here is the poet list: Geoffrey Chaucer (1340s-1400), William Shakespeare (1564-1616), Samuel Butler (1613-1680), Lord Byron (1788-1824), Walt Whitman (1819-1892), Emily Dickinson (1830-1886), T.S. Eliot (1888-1965), Allen Ginsberg (1926-1997), Sylvia Plath (1932-1963), and Dorothea Lasky (1978- ).

Without a direct comparison between real and AI poets, it's hard to claim the two are "indistinguishable."

Ok. Fine. Let's just ignore the "indistinguishable" part and more on to the "rated more favorably."

Then there's this plot.

Notice how the blue and orange bars are offset? Were they originally planning on providing a direct comparison between AI Whitman and human Whitman, but then removed that data for some reason?

Because the survey was of people who don't know poetry (90% have not read a poem in the past year), they probably pick up on the meter and rhyme and think, "yeah, that sounds good" and hence rate it more favorably.

That's fair enough. Who doesn't like meter and rhyme?

First, let's look at AI vs the human poetry it was trying to mimic.

I don't know who they are trying to fool, but these are clearly distinguishable.

But there is something really interesting about the AI poetry. It all has good meter and rhyme.

AI-generated poetry is indistinguishable from human-written poetry and is rated more favorably - Scientific Reports Scientific Reports - AI-generated poetry is indistinguishable from human-written poetry and is rated more favorably

Can I just rant about this paper really quick: www.nature.com/articles/s41...

It tries to claim that AI poetry is both "indistinguishable" and "rated more favorably" than human poetry, but it's methodology is... sus at best.

Let's talk.