Mastodawn

People keep assuring me that LLMs writing code is a revolution, that as long as we maintain sound engineering practices and tight code review they're actually extruding code fit for purpose in a fraction of the time it would take a human.

And every damned time, every damned time any of that code surfaces, like Anthropic's flagship offering just did, somehow it's exactly the pile of steaming technical debt and fifteen year old Stack Overflow snippets we were assured your careful oversight made sure it isn't.

Can someone please explain this to me? Is everyone but you simply prompting it wrong?

It's a good thing programmers aren't susceptible to hubris in any way, or this would have been so much worse.

Tom Renner 3h ago

Show thread

Bodil 1d ago

You know, it isn't even that tools like this are useless. There are absolutely things they could be good at. I've personally seen Claude find stupid little bugs you'd spend an hour figuring out and hating yourself for afterwards with great efficiency. I tried the first iteration of Copilot, back when it was just an aggressive autocomplete, and while I had to stop using it because it was overconfidently trying to finish my programs for me without being asked, it was great for filling in boilerplate and maybe even a couple lines of real code for the basic stuff. We have models nowadays that are actually trained to find bugs and security issues in code rather than having the entire internets thrown at them to produce something Altman & Amodei can sell to the gullible as AGI.

But there's the problem. The technology has been around for a while, we have a good idea of what it's good for and, more importantly, what it's not. "Our revolutionary expert system for finding bugs in your code" isn't nearly as marketable to the general public, and the CEO class especially, as "our revolutionary PhD level sentient AI that will solve all the world's problems if you only give us another couple trillion dollars, and also wants to be your girlfriend." And so we get Claude and ChatGPT and RAM shortages and AI psychosis and accelerated climate change instead of smaller, focused models that are actually good at their specialist subjects. Because those don't produce as much shareholder value.

Tom Renner 3d ago

GeePawHill 4d ago

One could spend all day every day telling folks using "mental activity words" to describe LLM output that they are wrong.

It does not "know", it does not "think", it does not "guess", it does not "figure out", it does not "reason", it does not "decide", it does not "feel", it does not "opine", it does not "believe", it does not "see", it does not "lie", it does not do *anything* you'd use a mental activity word for.

And every time they do it, anyway? They make the world just a little worse.

Tom Renner 4d ago

Emelia 👸🏻6d ago

Friendly reminder: last year I built FediMod FIRES, a protocol and reference server implementation for sharing moderation data.

I haven't yet been able to get anyone to adopt it or even signal intent to adopt. But regularly I see people complaining about the lack of data sharing when it comes to moderation, especially for combating spam, scams, and harassment. The tool is there, please use it!

Whilst I'm not actively working on FediMod FIRES this quarter, I did apply in November for a grant to continue that work, and last I heard a few weeks ago is that the grant made it to the next stage, so I may have some money again to fund development.

It's not 1.0.0 yet, because I decided it needed more work for me to be happy to call it that, but it is usable!

Installation is also super simple for data producers, literally two commands on debian or ubuntu boxes.

Learn more: https://fires.fedimod.org/manuals/reference-server/

#moderation #fediblock #administration

FIRES Reference Server | FediMod FIRES

Fediverse moderation Intelligence Replication Endpoint Server

FediMod FIRES

Tom Renner 5d ago

josef 6d ago

i go away for just a few weeks and now im getting Suggested Preprocessor Directives

Tom Renner 5d ago

Cat Hicks 5d ago

Change, Technically just hit over 25,000 downloads 🎉 🎉 🎉

I'm so proud of the level of rigor, research AND fun we've put into these episodes, a joy to make. A lot of work and a joy! Proud of us

@analog_ashley @danilo

Tom Renner Mar 23

Colin McMillen Mar 22

Allow me to introduce #MLL coding, the counterpart to #LLM vibe coding. MLL (Manual Labor of Love) coding allows one to spend more time doing a thing, and lets one get better, faster, and 100% understood code.

Tom Renner Mar 23

Show thread

Robin Mar 20

@trenner @hausgeist If you’re still talking about Codeberg, they provide Actions runners: https://codeberg.org/actions/meta

I was also looking for Forgejo Actions cloud hosting, but that doesn’t seem to exist yet and it’s incredibly easy to set up on e.g. a Hetzner host

Blog	https://tomrenner.com
Twitter (abandoned)	https://twitter.com/TRenner_
Github	https://github.com/tr325