

I work a lot with AI images, you just get a sense for it over time. It is getting harder over the years as things improve however.
Yeah that’s a decent way to start. Giving compliments is a great way to flirt! Saying something like “That was a great conversation, you’re easy to talk to” might even be a bit more flirty without putting yourself that much further out.
Just tell the US they can have any resources they can liberate from Russia. We’d have tanks at Moscow in hours.
It’s certainly better than most! For instance the text looks excellent. Look at the scientist’s eyes for a clue - one of them has a suspicious white circle while the other doesn’t, and the asymmetry does not seem to be intentional.
Looks like an AI generated image to me. Lots of strange artifacts an artist wouldn’t create. And there’s something uncanny about the stippling pattern I’ve seen before in AI images.
Star Trek’s WWIII happened from 2026-2053, I believe. So we’re still on track.
The thing I miss most from Reddit is all the niche technical communities. So much knowledge is contained in those.
Seems half-baked. Well unbaked really. They make a shit ton of assumptions that I’m not sure are true.
For example, why do they assume 90% pulverization efficiency of the basalt? Or is that a number they just pulled out of their ass?
And does ERW work if the pulverized rock is in a big pile on the sea floor? Or would we have to dig the highly radioactive area up and spread it around the surface?
And does the radioactive water truly stay at the site of the explosion? Or will it be spread through the entire ocean via currents?
Cool concept but, like, maybe we should check the assumptions a little harder?
Nuclear explosions are inherently unsafe…
…but fuck them fish!
Yeah the whole article is a press release used by Invoke (the company) to reassure their clients that they are legally safe creating things with AI.
Look, I think Invoke is good software, but they are being intentionally misleading to make money.
The open paper they published details the algorithms and techniques used to train it, and it’s been replicated by researchers already.
My comment was from 8 months ago…
They could reorganize as “NOT-U” under all the same terms but without the US.
It’s trivial compared to the compute they dedicate to AI models. Like, not even a rounding error.
The hell? There’s no reason to use plain HTTP instead of HTTPS.
And symmetric encryption is wildly irresponsible as well.
You can be vocal about it and still vote for the lesser evil in order to reduce harm and save lives.
Yeah I was being a little sarcastic. I agree that using a single lightweight framework is ideal.
Runtime js for rendering your page is still bad for performance of course. Best to use static html where possible.
I got bad news for you… htmx is written in JavaScript.
You must have a lot of memory, sounds like a lot of fun!