It's not fast.

you don't even need you prefered python version, uv will download it.

I can't wait for the machine to do all my work so I can finally do my personal projects without being interrupted.

I have so many personal projects that I've started over the years, and then left to wither on the vine. I've been able to complete a dozen or so over the last 2 years, and work on a handful consistently over that same period, using AI heavily, and it's a lot of fun. I can work on the high level ideas, create projects, spitball with various characters and simulations, and it's like having a team of digital minions and henchmen. There is fun to be had, and you can us AI well or poorly, so you can develop your own skills while playing with the systems.

There's still just something magical about speaking with a machine - "put the man's face from the first picture onto the cookie tin in the second picture, make sure he still looks like Santa!" You can have a vague idea or inkling about a thing, throw it at the AI, and you've got a soundingboard to refine your thoughts and chase down intuitions. I totally understand the frustration people are having, but at some point, you gotta put down the old tools and learn to use the new. You're only hurting yourself if you stay angry and frustrated with the new status quo.

Yeah, but about personal projects we're probably different. They don't always involve a computer and my joy is in the making, not in the completing. Wither on the vine is fine for me.

Now back to computing, since I've been doing this for 25 years as my main job and it's probably what you thought I had in mind:

> at some point, you gotta put down the old tools and learn to use the new

I have the habit of learning new tools out of curiosity and only keep the ones that actually solve problems I have. Over time I have kept some (example: dvcs) and ditched others I was told were the best thing since sliced bread (example: containers). So far, conversational AI has been very good at replacing google/stack overflow. But that's about it.

I'm sure I'll use more of this stuff as time goes by, but there is really no need to rush things. I'll let early adopters adopt and I'll harvest mature solutions in due time.

The only appropriate response to such a proposition.

> I don't see how this can be a replacement for javascript/XHR based autocomplete.

It can't do complex autocomplete. It's ok for simple cases only. I use it with a 25k long list to ease the input. Works well enough for this.

Ah... the old "all or nothing" fallacy, which in this case quickly leads to "save the planet, kill yourself". We need more nuance.

What is the nuance? Let us know.

What are his hobbies? Let’s pit them against training LLMs in value and pollution rate.

gremlins

Since kiddos play them at 1.5, you would actually get a normal video :-P

> it happened to look a bit better than Perl. Which is absurd to me.

It's not absurd, it's a debatable point. I think being easy to read is definitely a desirable quality for any language, especially for adoption. Some are better than others in this field.