"One note from a science process technicality: the closed-source nature of these LLMs makes reproducibility very difficult, which weakens the strength of their result."

No more so than a study in psychology.

That's the truth of it, and this is arguably the real news here. I've got agents that do more web browsing today on my computers than I do.

Eh. That's likely a small fraction of a fraction of a percent.

Look around you. People are not generally concerned with privacy, nevermind security.

English is a mishmash of several languages, so learning just English is enough.

It was J for me

It's not a matter of politically acceptable, it's something that doesn't alarm shareholders.

So let me get this straight. If I live in a hot or cold environment, and/or I regularly use the advertised range (or as much of it that the car will give me, realistically) to avoid having to sit and wait for fast charge to 80% for an hour on a road trip, I'm looking at (probably) no more than 80% capacity available after 5 years?

That torpedos the used market solidly. No wonder they're available so much cheaper.

This is the blandest, most watered-down way to say "lost their job" yet. Quite nauseating.

Just trying to keep people safe. Nothing to do with lock-in, rent seeking, control over everything you watch, or mass surveillance. Not about any of that AT ALL!

Hang on, let me Google it...

/s

I went to school in VA. There was rarely a day that the heat would get turned on. They're big monolithic heat islands full of children, calorie powered.

We also didn't have A/C back then, either. Windows would be open all day. I doubt they do that now.

I know this comes off trollish, but: why can't these power companies (which are usually public utilities) just build more power production instead of jacking prices? What's the deal? Why is this a "hate on AI that runs in datacenters" problem, and not a "just produce more power" problem?

It's the same in the US. I use direct deposit all the time to pay bills or other recurring payments at no cost.

Yep. The fundamental problem that requires loops is that opus et al are lazy AF. They do not "implement the plan, make no mistakes". They'll do a subset of {A..M} phases in a plan (90% of A, 70% B, 30% L, 0% M, etc.) and then say "all done!" when it compiles. So, you've got to loop it "do this until it's done". It's fundamentally brute forcing the problem, because the models aren't designed for completeness, just complete-enough, and then lies to you.

The harness exacerbates the problem. People have implemented some privately which do this correctly, but aside from the one I just made available on gh, I'm not aware of any that are public which do so natively/by core design. (And even then, it's sometimes iffy...)

This is all just marketing to try to cover for the fact that Claude Code wasn't properly conceived or designed on the onset to do what agentic tools like Hermes (and others, like Meept, or that Paperclip company with its autonomous employees) already do: create autonomous agentic workflows with clearly defined executors.

"It's a loop" is just bullshit to cover for the fact that they've got no clear, clean way to constrain context or workflows. They're trying to make themselves sound edgy so they can seem at the forefront of something they've clearly fallen far behind on.

Watch, they'll come out with some "new" feature in a couple of months which is already old hat to those at the forefront.