TechTakes

Starts with a good zing? That thread was murder start to finish.

My business has been releasing "AI" implementations for business since early days. The lack of understanding and just silly reporting on the subject matter is hilarious, disconcerting and sad.

So her entire rant starts by talking about AI safety, and then reduces the conversation to talking about AGI being created by text generation AI systems. I'm a bit confused, is she specifically just dunking on shitty reporters who only cover the "AGI from text generators" drivel?

EDIT: Alright, I finally got my half-asleep brain around it. I'm gonna leave my AI safety rant below anyways.

Like, AI safety is an actual problem right now.

• We're coming up on a period where fake AI-generated footage is actually believable at a glance.
• AI voice generation software is REALLY FUCKING good right now.
• We have drone systems which coordinate swarms well enough to perform intricate light shows.
• AI-generated art is also insanely good at the moment.
• newer facial recognition systems are scary good at identifying people now

These are each a problem because:

• forging video for whatever fucked up purpose is easier than ever. Revenge porn, twisting a political narrative, etc. All super easy now.
• again, forging audio. This combined with the video, the possibilities are endless
• The idea of digital-art-for-profit should pretty much be viewed in the same fashion as professional calligraphist, within the next couple of decades. Vast majority of digital art jobs are probably going to be dead. Animation is probably next on the chopping block.
• I don't think I need to go too in-depth on why AI-controlled drones looks really good for military prospects.
• Mass surveillance that's borderline impossible to escape is on the horizon, thanks to AI.

Like, I don't see how this person can shit on AI safety while completely ignoring the actual vast majority of AI safety issues.