yoz-y

Assuming the company still gets some value from them. France has good worker protections but you can always just close shop. (And move elsewhere).

Another optimistic look is that there might just be more work, that otherwise would not be done.

For example we always complain about AAA games coming out riddled with bugs, now we have an opportunity to cut a significant portion of that if we also keep the workforce. Same with administrative tasks that take forever because it’s one person filing it for the whole country etc.

The purpose of civilisation (or rather, why it exists in first place) is that “civilized” peoples managed to survive longer and reproduce more and produce better means to take over the land of those who didn’t adopt technology.

Now we are going to see if there is a limit or not. Or if AI changes the game (the dystopian end game is a small amount of people controlling all the production and getting rid of the rest by one way or another)

I’ve had good experience coding with it. In a bit of an ironical twist I now find Gemini a much better coder and Claude a much better search engine.

Interesting. I guess I’ll have to go with the m5. On one hand it feels bad to buy older generation hardware. But then. It’s already so much faster than what I have…

I am due to an upgrade (still using the 2019 intel). However I am torn between a M5 Pro with 48 Gigs of ram or waiting for M6 pro. I’d really like to run local models for once.

I am doing a lot of LLM writing and porting; and am still calling it slop. Even though it saves me a bunch of time at work and in hobby coding.

To me the main difference is that there is no staging (so no stashing needed), and conflicts are first class citizens. In practice this means that you can usually halt your work halfway, switch a commit, rebase everything onto head and continue working even if one of the branches off your commit has conflicts. In git you would have to resolve those before the rebase can finish.

Since conflict resolution these days mostly boils down to “hey LLM, resolve conflicts” this is much nicer.

Also jj tracks every change to the repo so undo works much better and is more readable than reflog

TIL that Broadcom is a 1.6T company.

In a way it’s sad that AI came in the age where a lot of us are jaded because the internet which we built has become a place for gambling and scamming. I remember when a new language or framework was met with excitement. Today I hear (and am a source of) mostly criticism.

Yeah. If you can’t do something that’s fine. I draw as a hobby but I suck at it. When I see some of the pre-post autotune performances I can’t feel that people are just feeling entitled to a paycheck.

It does right? On the other hand I don’t believe that AI will reach the hypothetical perfection. Not in the state it is in now.

To me a lot of similar games are fun for 5 minutes. But they become rote after a couple of hours. One could say that it’s nice to have spent a few hours having fun, others would get a feeling of regret.

If the AI gets to the hypothetical perfection we will share precisely none of experiences.

Teachers get paid?

Trainers, coaches, therapists… yes.

Am not. Recently I started reading more classic literature and realized just how much my comprehension is lacking, even after working in English exclusively for years.

This is actually useful. I didn’t know that generic “you” meant “one”. I always interpreted it as “you the reader” and felt desperate once a week when reading articles like these.

Would be nice if articles defined the requirements to be a “you”.

There was the “sweet solution” :) When reading the stories about how the first iPhone came to be, had they tried to also have an sdk with the first release, it would not have shipped on time. And although Jobs was famously stubborn, this is something he changed his mind quite quickly on. In no small part due to the ecosystem that popped up despite him thanks to jailbreaking.

Never tried it myself but I’d be afraid of stuff being randomly broken. Eg at some point pi-hole made YouTube history not work.

Moving from IE to Firefox was a much more memorable experience than going from Firefox to Chrome. But yeah, first Chrome was fast. I’m quite sure any of the today’s browsers would be fast if running in the pre-react world though.

Problem is that for any non trivial software is that the prompt is a conversation and the steps increasingly depend on the output. Of course you can ask your LLM to write the SPEC.md at any moment but it will lack nuance.

Shameless plug: at https://slopyard.net I embrace this and only let people submit the spec. If somebody wants your software they can regenerate it and change whatever they want.

And so the game of cat and mouse goes on. Your point being?

Excited to finally see this out. When new big models come out I like to make them audit any other code I generated to see how things are.

Unfortunately there is no way for us to try this. “yes” might very well be the answer. My hunch is that although it would seem pointless, you would still be able to go through the motions since the work is largely physical. For knowledge work, concentration and perseverance is half the work itself.

That is not an apt comparison. Imagine you were a hunter and were given access to a supermarket for 10 minutes.

Would you become unable to hunt?

And if you’re in the middle?

I’ve decided to build my alternative because every time there is a prosumer tool I like (Pixelmator, affinity) they get bought out (affinity to canva), or the developers suck (Pixelmator developers cramming all features people wanted into a different “pro” version)

More info in another comment. One of the reasons I’m doing this is to also see how (and if) it is actually possible to build a complex software that I can use.

Tiny tools I’ve made tons, but something that needs actual architecting is different. Image editing is also a middle ground candidate. The output is what matters and is the final test. “Subtle” bugs will not ruin your day as long as you use a periodic backup

So. I’m vibeslopping my photoshop too.

Not yet production ready although working pretty well.

I based it off babl, GEGL and MyPaint. It is binary compatible with gimp through xcf and also supports svg natively.

Since I’m a Mac user it uses metal rendering for speedy painting, gradient and filter previews, then commits changes through GEGL asynchronously. It does work quite neatly.

It already found a couple of bugs in GEGL and Gimp, which I have tucked in patches and tickets. My plan is to go through these manually and do contribute the fixes, if the community accepts them (and if they’re indeed problems) However I don’t think an LLM should send patches.

I disagree. For me this shows that people can fall in love with a non-human, and being human is unnecessary for the purpose of giving (some) therapy advice.

LLMs do sound human, I’m not contesting that. But if pressed they will not react as humans. There is an infinite amount of questions that an LLM will answer for which a human would give you a blank stare, told you to shove it or call the police.

The conversation varies between “can’t be” and “isn’t”.

Most sci-fi AIs are way beyond our current models. I’ve yet to see one that would actually pass a Turing test (that said, I’ve not tried models that were post trained to actually try to behave like a human).

Consciousness itself is a term that is not really defined anyway, so depending on each’s take the answer will be different.

Yeah, shopping at decathlon feels like stealing honestly. Uniqlo has them too iirc