Post

I despise AI companies and their goals. I am deeply passionate about climate action, and I weigh any positives that AI can bring against the critical need for us to save our planet right now. I am passionate about music and art, and I do not want it gone forever. But I am also not a dick about it!

I think that the fringes of the anti-AI movement have lost all context about how widespread these tools already are in society. The cat is already out of the bag. I see it every day with my students, fellow educators, or friends in software engineering.

We need to work on tight regulations, laws that protect human work, protecting living standards, and continuing to fight the climate crisis. But AI in coding will never go away at this point, like how LLMs applied to machine translation have been doing just fine for ~10 years on sites like DeepL.

Maybe you should google "winning hearts and minds" and see how that strategy has gone, historically. Then you can stop with this nazi-friendly bullshit and just go to war on behalf of not being evil. People bitching about manners aren't going to be won over, they just want to avoid social costs.

Using an LLM doesn't make you a Nazi, which is just as well, because otherwise we'd have over a billion Nazis out there. The AI bad guys aren't on Bluesky.

Being a data center cheerleader and "using LLMs" aren't synonyms in my mind...

The “AI” is two futures: “centralized AI” where everyone is a serf to a few players running it. And ”AI for everyone“, where everyone can host an AI locally. I am very much against the first trajectory, despite being ”pro-AI”. Just a data point-maybe there‘s a bit more dimensionality to this debate.

The problem is the very real risk of knowledge/intellectual property losing value faster than societies can compensate. With AI making it super easy to quickly copy and automate whole businesses and business models, the only ones able to compete and make money will be people able to outspend others.

In that scenario, which we’re heading towards IMO, locally-run models won’t make much of a difference, because no smaller- to medium-sized company will be able to compete with products like "Claude for XYZ" that will be backed by advertising and lobbying budgets in the dozens of millions.

I think that this is something that a lot of the anti-AI crowd miss by choosing to bury their heads in the sand. Software developers will be (and already are!) the first to try and make this technology not contingent on trillion-$ companies, already with huge success. I'm hoping for option 2 too.

It’s a weirdly well trained reflex to reduce the complex issues to “us vs them“ mud-slinging (with “them” taking various values - from “AI“ to “immigrants“ to “wind turbines” to “data centers“) and get outraged, and we observe its effects lot now.