Historically, AI has found and used exploits. Before OpenAI was known for chatgpt, they did a lot of work in reinforcement learning (often deployed in game-like scenarios). One of the more mainstream training strategies (pioneered at OpenAI) played sonic and would exploit bugs in the game, for example.

The compute used for these strategies are pretty high though. Even crafting a diamond in Minecraft can require playing for hundreds of millions of steps, and even then, AI might not constantly reach their goal. Theres still interesting work in the space, but sadly LLMs have sucked up a lot of the R&D resources.

The challenge is that AI for a video game (even one fixed game) is very problem specific and there’s no generalized approach/kit for developing AI for games. So while there’s research showing AI can play games, it’s involved lots of iteration and AI expertise. Thats obviously a large barrier for any video game and that doesn’t even touch the compute requirements.

There’s also the problem of making AI players fun. Too easy and they’re boring, too hard and they’re frustrating. Expert level AI can perform at expert level, which wouldn’t be fun for the average player. Striking the right difficulty balance isn’t easy or obvious.

You gain brouzouf…

Is there enough gear/experience to just skip the open world stuff? It wasn’t clear to me when playing if I would hit a wall and needed to grind on the open world to progress.

Yeah your understanding about the towers is correct. I don’t think it’s inherently bad, I’ve even enjoyed it in some of the AC games, but in rebirth it just feels like a bad chore list. Some of the combat challenges can be interesting but the ones with the summon stones (I forget the name, but they reduce the power of summon fights and do other things) and scanning the life springs are just terrible imo.

This is a pretty bad take and I feel like you must have not really played tribes very much. Comparing tribes to CS? Really? You think the popularity is because of the team sizes?

The fact that they are only implementing 16v16 seems like a warning flag to me. I wouldn’t be surprised if this ends up as a mediocre experience with a few tribes mechanics just largely trying to cash in on name recognition.

Yeah I’m sure shell and co really forced the sale of 750k F-series trucks last year, right?

YouTube already does that, though? In fact your premium sub helps creators much more than a non-premium viewer does. Of course the multiplier proposition would be new/different.

Not trying to shill for YouTube but that's one of the main reasons I have a subscription is to support creators I like.

Basic features like turning a bulb on with a set brightness or color (if you’re lucky, you can turn on with a set brightness). No support for scenes (due to matter spec) which can lead to issues sending commands to light groups. And if you want to actually update bulbs good luck because it’s buggy as hell.

I wouldn’t say they’re great; they’re adequate. There’s a lot of issues with the bulb firmware that still needs ironing out both with stability and feature set.

Liking a game despite its flaws doesn’t mean it’s a good game that deserves a high score.

I’m not really convinced that a GPU backend is needed. Was there ever a comparison of the different CLIP model variants? Or a graph optimized / quantized ONNX version?

I think the proposed solution makes a lot of sense for the task at hand if it were integrated on the pic-rs end, but it would be worth investigating further improvements if it were on the lemmy server end.