This is just wrong. I just tried GPT-6.2 Astra. Here's a 1 shot result trying to improve a game scene
My prompt: "The NPC in this sequence feels fake. Please make him do realistic gestures that you would see when people speak in real life & any other improvements"
It nailed it:
- Eye contact: much less of it. This NPC hates the main character, so he barely looks at him now
- Hand gestures: this NPC fucking hates his job, so he is crumping to the 5 stages of grief
- Positional Audio: when a character talks, their voice should move where their mouth goes. So when he turns around, the speech gets quieter and more muffled
The only thing it missed that would have made it AAA+ quality would be realistic props & clothing, as it's raining a lot. Dual lightsabers & a bikini would bring the whole experience together
Theo - t3.gg @theo Nov 17 Why I hate GPT-6.2 Astra During early testing, I found myself so frustrated with this model that I spent a day writing a letter to OpenAI about it. They fixed a small handful of my complaints, but not enough for me to
People keep telling me this is fake & I don't even have access to Astra
Again, lies and slander. Here's a screenshot of my GPT Work homepage. I tried the left suggestion card, and nothing more
You are simply jealous of the results!
This is just wrong. I just tried GPT-6.2 Astra. Here's a 1 shot result trying to improve a game scene
My prompt: "The NPC in this sequence feels fake. Please make him do realistic gestures that you would see when people speak in real life & any other improvements"
It nailed it:
- Eye contact: much less of it. This NPC hates the main character, so he barely looks at him now
- Hand gestures: this NPC fucking hates his job, so he is crumping to the 5 stages of grief
- Positional Audio: when a character talks, their voice should move where their mouth goes. So when he turns around, the speech gets quieter and more muffled
The only thing it missed that would have made it AAA+ quality would be realistic props & clothing, as it's raining a lot. Dual lightsabers & a bikini would bring the whole experience together
The graphics programmer behind the water technology in #NoMansSky is giving a talk today titled:
"The Changing Tides of NMS: Crying Saltwater Tears Trying to Implement Dynamic Water on a %@$#&ing Sphere"
Good luck, Tom!
graphicsprogrammingconferenc…
If your architecture is good, this is nonsense advice. Just get pen tested and have LLMs audit your codebase
Your codebase is really that bad? Ok, use LLMs to rebuild it - with capable developers. Non-technicals will be gaslit by an LLM into a bigger mess
This isn't really my opinion, either. It's basic common sense. Unless, that is, you think LLMs write better code architecture than humans - I've seen juniors write better code. They're a powerhouse with LLM
If you do think LLMs architect better than most devs, it's a skill issue. Sorry. I've worked with hundreds of devs, read an inordinate amount of open source code. It's simply untrue
LLMs able to find vulnerabilities on mass != able to put together maintainable architecture. For the same reason AI comments are obvious: you can't computationally verify "good sentence structure". It also requires long term consequential reasoning (not the same as test-time compute) and memetic social context.
Other than that, it's like tearing down a perfectly good building that has no security cameras or alarms, to then rebuild it worse but adding security cameras and alarms. A good way to waste a few million $
If the building is crumbling, agree, but it'd be in your best interest to sit down with an unbiased party before making that decision
The lack of context here is irritating but I get @chamath is broadcasting to a specific audience. It is worded in an absolutist sense though, with a lot of unfounded implications
All, I repeat ALL, existing software needs to be rewritten. Because this software was written by humans it is both inefficient and insecure. The cost of inefficiency and insecurity in a pre-AI world was moderate, but in a post-AI world it will destroy your company.
This new effort should not be a bottoms up exercise where engineers get to have an opinion about how they set up their repo or which coding assistant to use.
This needs to be done in a new, top down system of record that focuses on auditability, drift detection and verification among other things.
Every company needs to do this asap as today has shown.
If you want help: 8090.ai
3 hours for that, by the way. 3 flipping hours of an LLM trying to make sense of natural human movement
If I seriously dive back into this, thank GOD for GVHMR & the latest markerless mocap tools. I'll need them!!
I gave Opus 5.5 the task of tidying up the intro
Previously the estate agents hands were clipping through the wheel & his bones broke while closing the door
So, it added IK to the fingers?? And his bones still break
And now his mouth & arms move when he talks.. just watch 😂
And I'm out of the @UnrealEngine MegaJam!
I got too ambitious, so:
- Everything left at 70% done
- Including the main game loop. Which meant it felt confusing and disjointed
Every failure is a lesson! Let me know if this is worth me polshing + releasing
5 mins of gameplay:
I appreciate all the love on this post!
I had several DMs too, more (legit) ones than I ever had in a day
This deserves to be finished, but I have other commitments so it depends
Regardless, I am happy. Even if I missed the deadline! It was a fun excuse to try @UnrealEngine
And I'm out of the @UnrealEngine MegaJam!
I got too ambitious, so:
- Everything left at 70% done
- Including the main game loop. Which meant it felt confusing and disjointed
Every failure is a lesson! Let me know if this is worth me polshing + releasing
5 mins of gameplay:
Oh and perhaps my biggest blunder, I had to change the whole script in the last few hours & redo the game sequence.
It felt quite good, but then I picked up on the "13+" age requirement - so I removed all the gory bits & shifted it around so it didn't feel off
I gotta learn to read the rules before I dive headfirst into ts haha
I definitely bit off more than I could chew, but it lead to me realising I can get waaaay more done in 3.5 days than I ever thought possible
This doesn't look half bad, and with voice acting, manual level & sequence editing, it'd probably be an experience worth releasing!
And I'm out of the @UnrealEngine MegaJam!
I got too ambitious, so:
- Everything left at 70% done
- Including the main game loop. Which meant it felt confusing and disjointed
Every failure is a lesson! Let me know if this is worth me polshing + releasing
5 mins of gameplay:
Uff I've had an hour to play test. It is so janky
I know how to get it where it needs to go, but in 3 hours with time to build and upload? Not sure guys.
I'll put another 1hr 30 into it and see if I can fix the entire inner loop gamefeel in record time 🤣🤣