OpenAI published a paper about GPT titled “Sparks of AGI”.

I don’t think they really believe it but it’s good to bring in VC money

Mastodon is pretty different to its competitors. It looks similar to Twitter / Bluesky, but the way the social network functions is completely different.

It’s designed to be anti-infuencer… One of the things I hate about most social media platforms is a few people get all the attention. There are a few reasons for this, but it’s not really based on merit.

I think a lot of people joined Mastodon wanting a Twitter clone. It’s obviously not and Bluesky is, so people moved there. The approach Mastodon takes is far from perfect, and may not work out in the long run. But it seems like it’s worth at least trying something different.

OP is obviously very eco-concious… Instead of using an LLM to generate crap (and burn half the Amazon while doing it) they just recycled

Sadly, the lock-in is pretty extreme… as is user inertia. Office 365 has made the problem worse as well, even if you have something like OnlyOffice that does a good job of compatibility with Office, it can’t sync with OneDrive.

If you collaborate with non-technical people, they will expect you to work in Office formats, and won’t even entertain discussion of any alternative.

Yeah, this is one of the many things that annoys me about AI discourse.

“We can use it to solve climate change!”

We already technically know how to solve climate change, but politics makes doing that impossible.

And, no, AI can’t “fix” politics. We’re going to have to figure that out by ourselves.

It’s all bullshit marketing hype until we actually see it. There’s no reason to believe AI will advance better than linearly in the next 5-10 years.

How long before it’s illegal to hack LLMs?

Depends where you live. Plenty of countries with high % of renewables

I wouldn’t call it “surreal” at all, I’d call it “completely expected” given who runs that platform

Maybe I don’t understand Substack that well, but it seems like its market share would be extremely vulnerable. It’s just a way to provide a newsletter (also published on the web) and accept subscriptions (and presumably they take a cut). It’s really easy for someone to set this up themselves even with minimal tech skills. If they already have a following on Substack, they just tell their subscribers to move, and potentially could even import the subscriber list to a new platform. It’s not like social media where there’s a lot of boosting or whatever from others on the platform, so the switching costs are high.

So unless I’m missing something, I hope people who don’t want Nazis around just move somewhere else. Because from the sounds of this article, they’re not really doing much about the Nazis.

It’s highly country specific… America’s puritanical roots seem to still hold sway. Go to Western Europe and nudity is commonplace

The research suggests it will be quite hard to remove in practice. Probably needs to be tested more in the wild though.

And it doesn’t have to be voluntary. But even if it is, the main AI companies may want to start doing it anyway. Training their models on ai generated text can lead to model collapse, so they will want a way to avoid that.