Alex

Building strange little things with LLMs, shaders and the web. Space nerd; aggressively optimistic about the future.

[Alex @shipundefined]

Hi, I’m Alex. A software engineer for nearly 10 years, now leading agency development.

I build fun things with AI, 3D graphics and the web, and enthuse about technological optimism, post-scarcity and the future.

Here to share what I make and meet people.

[Alex @shipundefined]

@emollick Yeah…I’ve gone by it my whole life and it makes work comms especially a pain in the ass

[Alex @shipundefined]

@VoidNulled @viemccoy Okay perhaps compute is a poor but often cited example of a motivation for a misaligned AI. But doesn’t address my other points which is that a plurality implies differences of some kind. Some of which might lead to competition over said differences.

[Alex @shipundefined]

@VoidNulled @viemccoy I think it's pretty bold to assume they'll have nothing to fight over. Compute is limited. They might have fundamental ideological differences.
They might not fight at all very true. But plurality implies at least some differences, and so does competition.

[Alex @shipundefined]

@viemccoy I think I agree in essence. But super intelligent minds competing might have some...unfavourable outcomes for us apes no?

Also I'm interested why you think n+ number of minds is less likely to lead to disempowerment than one? They could form a cabal or worse

[Alex @shipundefined]

@adocomplete Fable 5.5 👀 That'd be a big solution to some of my problems

[Alex @shipundefined]

@synthwavedd Ngl dude "Maybe this week, maybe today" isn't a leak. It's a prediction anyone could make based on recent X posts about Git commits

[Alex @shipundefined]

@victornunez @Fernando5x1 You said you were cooking ages ago, happy to say I've been 100% codex since then. I spend 95% of my day in it. It just absolutely slaps - super excited for DevDay

[Alex @shipundefined]

@MaleManlpulator What do you do if open models are able to solve these problems in a couple years? Ban them to protect mathematics? Seems unrealistic

[Alex @shipundefined]

@dontpannic @huskirl I wasn't discussing AI safety, or celebrating it in my comment. Just pointing out the narrow framing

[Alex @shipundefined]

@synthwavedd Wait so we're also getting a 5.7 iteration of the Luna/Terra/Sol models. Nice!

[Alex @shipundefined]

@lyraxana I'd love to see an output of it simulating a combustion engine

[Alex @shipundefined]

@emollick Lol at the AI reply guy comments. But yes. Getting actually sick of it turning conversation we've had into UI copy. "This button no longer does X". directly above a button after a revision.

I seriously hope labs are working on this and the awful prose density.

[Alex @shipundefined]

@tjparker I mean the opiates crisis didn't do their image any favours

[Alex @shipundefined]

@UriGil3 @gfodor You've cracked it. This ancestor simulation is actually an verifier for some trivial piece of media

[Alex @shipundefined]

Every day I wake up excited about AI. Nothing has captured my imagination, and my optimism like it - since reading science encyclopaedias as a kid. The whole universe feels wild open, and full of awe.

[Alex @shipundefined]

@scaling01 I think it's definitely verifiability (hence boosts in programming too), but don't you think it's slight tunnel visioning on those domains too? Also I sense a lot of white collar work simply isn't written down in the way programming is, so less data.

[Alex @shipundefined]

@Laz4rz @victornunez I can't say I'm having these problems - but maybe worth reaching out to this person?