Huh? You think the public versions of the models have been “lobotomized” so they don’t know how to turn off a PC?

It’s not lobotomized, it’s a simple harness restriction that has nothing to do with the model or its capabilities. And either way, I’m not sure what that has to do with “negligence” or “intent”? You think frontier labs should be prosecuted because they don’t allow agents to turn off your PC?

I don’t see the analogy? AI is not preventing the craftsman programmer from coding. Your compiler hasn’t stopped “yielding fruit”, so to speak.

The correct analogy would be, your mom can also just go to the supermarket and get much cheaper mass-produced tomatoes; which is… exactly the current situation?

I agree. If grandma grows her own tomatoes she can only try four tomato-related recipes. If she goes to the grocery store she can try thirty.

The AI is enabling a lot more experimentation with software. The AI supplies the tomatoes (infrastructure / grunt work), grandma tries her recipes (the software / business idea).

26 extra recipes are irrelevant if the person finds far more joy in the process of growing the tomatoes than in what they cook with them.

Then do that? At the very least, the presence of more code(/recipes) does not hinder an individual trying to do it themselves.

Modify the analogy: nobody cares about the mother’s tomatoes and everyone around her says they’re indistinguishable from the cheap supermarket variety. Even if it’s not true

That’s a fair analogy as well, though it makes a totally different point.

I think you’re right about the analogy itself, though it’s far from clear that society owes it to your grandma to hinder everyone’s access to cheap tomatoes just so she can feel more useful for her tomato-growing abilities.

I haven't seen people say it's indistinguishable quality, but that it doesn't matter. Quite similar to mass-produced foods in the supermarket, actually. We all know they're not as good as a small batch made by hand by someone, but they still get the job done.

Uh, no? It obviously matters.

The market for small batch locally made produce is huge. Entire grocery stores dedicated to local organic offerings is the standard in every upper middle class neighborhood on the entire west coast. The reason is obvious: a salad made with crappy tomatoes is a crappy salad.

We aren’t at this point in software yet because most consumers of software have no taste

“Moves to advisory role” is just corporate speak for “retired/fired”. I can’t think of a single example of an executive who “moved to an advisory role” and demonstrated even an iota of influence after that point.

Right, that's corporate speak. Welcome to gov speak:

> A Special Government Employee (SGE) can perform temporary federal duties for up to 130 days within any 365-consecutive-day period

https://www.flra.gov/Ethics_Rules_for_SGE

As an SGE, the person has legal influence, but ethics rules are relaxed. As an "advisor," there are almost zero ethics rules, and their influence is not legal, but wink wink. Many of the most influential people in the current gov are "advisors."

Considering that the very first paragraph of their README has:

> toast is for the developer who wants to stay in their terminal without spending hours getting it up to snuff.

I think he knows you can get anything you want in emacs/vim. The pitch is that toast is batteries-included.

Mouse support and filetrees are included in both vim and emacs. The batteries are already included and have been for essentially ever

Plenty of easy configs for them too. Lazyvim has been pretty batteries included and easy to modify for me and I'm definitely the type that did not read much of the docs

What on earth would you do with an iOS device that has 96GB of RAM? You realize the last generation (i.e, the most powerful consumer phone in the world today) had 12 GB of RAM? You feel the need for eight times that amount?

Number go up, I guess? A lot of people derive great satisfaction from ever larger (or sometimes smaller) number. But, uh, 8X is probably too much, need to spread that out over a few generations.

They obviously want to run AI models on their phone.

Very obviously you get a Mac Studio for that, not a phone. (Oh, the 96GB Studio is almost $6000, you counter? Imagine how much it would cost in a phone form factor).

You can't possibly expect Apple to take that market segment seriously.

In theory it sounds alright, although in reality, the heat and power consumption make this less than ideal.

I am not sure single digit tk/s on weak model like ChatGPT OSS 20b @ Q3_K_M is going to be worthwhile.

Not to mention this has to be <0.01% of users who "want" to do this. The iPhone tooling would make it hard to even configure and spin up a local LLM.

> the most powerful consumer phone in the world today

According to what? (I don't disagree, just curious)

I don’t know too much about image processing, but my intuition tells me that distinguishing a photo of a 3d scene from a photo of a photo of a 3d scene is a much easier engineering problem than distinguishing it from an AI-generated diffusion. So even just reducing the problem to the former is still a worthy accomplishment.

Then you won’t be able to prove it was a genuine photograph? That sounds like the right failure mode.

So this site is just a fan-fiction that thinks it's somehow dunking on Claude? I've never had a session that remotely resembles any of this. I honestly can't tell what point this site thinks it's making.

So you are lucky, congratulations.

Maybe, if you tried, you could concoct some adversarial example of a stylesheet that a recent Claude model would trip up on and fail to color a button correctly on the first or even second try. But it would have to be some explicitly engineered trick, akin to an optical illusion for humans.

You can’t convince me that I’m somehow the odd one out because I regularly have no trouble getting Claude to recolor buttons.

I wrote my own harness to stop shit like this from getting to my attention out of frustration.

I'm sure there's quite a bit of variation from person to person in these sorts of experiences, based on your harness, the way you talk, the stored memory, your CLAUDE.md, etc. But people absolutely have had this Opus 5 style experience the app simulates.

That’s a skill issue.

Will somebody please let Professor Tao know that he's simply experiencing a skill issue?

I'm so curious if you're actually able to keep a straight face while typing that.