You know the thing where AI companies describe their alignment direction as “Honest, Harmless, Helpful”?

The only one that’s marketable to the idiots buying AI services is Helpful. So that’s always top priority even when it goes against the other two.

Average decider wouldn’t stand an “assistant” that sometimes answers “I don’t know”. So instead agent providers train their already hallucination-prone model to be 100% confident in whatever it says. Who cares if it’s true if it looks like it is?

Your username with that conclusion paints a very funny picture.

I miss the time when sunglasses made you look cool. Fuck you, Zuckerberg.

It’s pretty obvious when they all seem to be competing for the scariest “containment breach”.

Anthropic keep posting article after article about new ways their LLMs keep “surprising” them with unexpected behaviour. This year’s “uncontrolled, autonomous” hack news from every AI company are completely absurd boasts. In normal times, everyone would be wondering why they sound like they’re proud of obvious incompetence.

But since we’re in super-hype “keep the money coming, AGI is tomorrow” times, instead, those are used to sell the lie of mysterious superintelligence. Be enthusiastic, be scared, we don’t care, talk about it!

I don’t design web apps but I’ve heard from some that do that firefox occasionally has problems with them, because it relies a lot on caching (actually they used the term “aggressive caching”, with a few extra choice words).

Anyway, usually not a problem. I use edge only for work since it’s the default browser and I barely do any actual browsing with it, mostly light app clients. On my own devices, it’s been firefox only for years.

Still a good thing to keep documenting the many reasons why people shouldn’t use it in my book.

I was curious how you were supposed to configure the thing’s output… and of course it has to be a mobile app.

I’d prefer physical buttons I can check and change whenever like an actual old school power adapter.

Twitter deserves the mockery. And this attempt too.

I’ve seen AI summaries that looked like the AI had not read the article.

Notice the ugly-ass monkey pic in the background? That user made one other post, and it was an ad for some presumably shit app with this icon. That’s all there is to this.

“I just watched movie. It sucked!”

“Beep boop. According to trustworthy, verifiable sources, movie does not suck. Sanction : immediate ban.”

Wow you were not kidding. Some of them are extra awkward and unnatural too, it’s almost a parody of what slop text looks like.

We believe people should have more control over what they see — and more transparency about the information in front of them.

An early illustration of the idea — you decide how much low-trust noise reaches your feed:

Hey, they scanned that place for a SteamVR home world. It might be another wall, bit I think if you try clipping through it there’s a Portal turret that will sound an alarm on the other side.

It used the be, first page has what you’re looking for, and any result beyond second page is probably irrelevant.

Today it’s : the first two pages are the same mass produced slop copying from itself on dozens of shady domains and blatant ads… and maybe the third page has one relevant result from a real source. Maybe.

And probably even more ads sent directly to your eyeballs. If they’re not there yet they’re already on the roadmap.

Three CPUs with different architectures on one device sounds crazy. That’s amazing.

That’s a silly hypothesis. That kind of numbers only exists for jobs I am absolutely not equipped for (like, I have some level of empathy).

But I did choose a comparatively less stressful job with pretty good benefits, and I knew looking for an equivalent job in big corporate would get me at least 50% more in pay. Not regretting any of it (especially now that this job market is all over the place).

I get this, but as an aside I am a bit amazed that it takes some people converting everything into monetary value to see this.

IMO having healthy life balance should be the goal, my mental health doesn’t have a price tag someone can just pay for and take from me.

I agree.

Also. because of the way they work and how people use them as a source of information, all LLMs are dangerous.

The article says they’re supposed to be an alternative to fully robotic devices, with lower power consumption since their movement is… bug-powered I guess, but I am really curious what kind of fine control you can have on a cockroach.