This is efficient. It is trained narrowly and specifically, reducing electrical waste. It does not replace meteorologists because it relies on them to provide data and interpret the results. Not everything with AI in the name is bad and AI is not ChatGPT, that's just Sam Altman's marketing tactic.

I feel very strongly that a lot of the barriers to human innovation would be overcome if people could afford to chase passion projects, but that’s perhaps too socialist of me for the mainstream.

+ reproductions of human work, and telling people how to off themselves.”

I could probably cohere all this better but it’s very late and I need to get up in the morning. Last thing—I feel iffy about this coming from Vox given their track record with their unionized workers from what I remember.

But I cannot get behind this framing and I cannot take it in good faith. “Silly humans hate AI when it’s the thing that can save them” feels disingenuous when the reality is closer to “People are sick of the unmitigated growth of AGI after it’s being shoved down their throats, making subpar +

A better way to frame this argument could be “People hate AI. These models might save its reputation” or something other than an allusion to the replacement of human innovation and creativity yet again. AI is more than a slop generation machine and it’s incredibly helpful for specific use cases.+

I’m not gonna go deeper into the specific types of AI discussed in the article, because I need to do more research myself to grasp their impact on the industry. But I do know AI is a buzzword right now, and that gets clicks, and that’s why the headline is how it is. +

+ Love to do work on. But if the positions aren’t available and don’t pay enough to live off of, people aren’t going to do them.

(And arguably the birth rate decreasing is also a result of the increase in the cost of living! No one I know can afford to have children right now!)

+ leaving out the fact that we’re in a major affordability crisis at the same time entire disciplines are getting cut off from federal funding (like CANCER RESEARCH) and out of institutions. I know scientists and researchers in my personal life. Every single one of them has a Thing they would +

Are there enough research positions available for people who are in the field? Do they pay a living wagw or salary? Worth the amount of work one has to put in? Are institutions funding those programs? Is the government? Walsh briefly mentions lack of R&D funding as another factor but I think +

Walsh mentions immigration restrictions by the admin, which, yeah. Credit where it’s due, that’s going to massively impact the research culture of the US and where research gets done and also what research gets done. But there are so many other reasons that affect that too! +

+ (and the fairly blatant “there aren’t enough white babies” stuff coming out of the current admin) I really can’t take birth rate/population replacement seriously. So setting that aside: what ELSE might be a reason for this?

+ This further begs the question: why is that? And then Walsh talks about birth rates declining in wealthy countries and the overall population going stagnant and I don’t know about you but that’s a red flag to me! Especially with the rise in birthrate panic from the American Right Wing +

+ of the entire article is fraught. “We’re running out of ideas” in the headline leads to the question of why (and also how). Bryan Walsh argues that it’s because the amount of researchers in the field cannot sustain the level of innovation necessary to make the line go up, or even stay stable. +

I don’t know enough about machine learning in the hard sciences to really say whether things like AlphaFold are actually useful without any detrimental effects to the humans who work with them and benefit from their discoveries. That’s something to look into. I do, however, think the framing +

I do think the usage of “artificial intelligence” regarding things like LLMs and generative AI as well as stuff like AlphaFold does AlphaFold a disservice. I would be much more excited about AI if it was associated with tools like that, versus the aforementioned slop machines.

Examples given are AlphaFold’s protein structure prediction model and similar models trained to predict and evaluate organic and inorganic structures for thing like stability and compatibility. That seems pretty cool and very much a time saver in the research process, and I’m curious about it.

I had to tab out of the divinity one this year, which is the first time I’ve had to do that about a game trailer. It was…something for sure.