Right, we have not even solved games in the non-real time environment, so why bother adding additional constraints like “on real hardware in real time”. This is exactly like Tesla trying to switch from LIDAR to cameras before self driving is even solved. It’s avoiding the real harder challenge and going off on inane tangents. John is essentially bike shedding.
In this case, John is going off on this inane tangent because of his prior experience with hardware and video games instead of challenging himself to solve the actual hard and open problems.
I’m going to predict how this plays out for the inevitable screenshot in one to two years. John picks some existing RL algo and optimizes it to run in real time on real hardware. While he’s doing this the field moves on to better and new algorithms and architectures. John finally achieves his goal and posts a vid of some (now ancient) RL algo playing some Atari game in real time. Everyone says “neat” and moves on. John gets to feel validated yet all his work is completely useless.
I understand that. Doing games in real time is just a performance problem that can be solved with more compute or inane optimizations. It’s not interesting research.
Surprised these people had to iterate and run these experiments. I thought all of this was common knowledge in books. Maybe it was an experience before this was common knowledge, but it’s not uncommon to see “painfully figure something out instead of cracking open a book, write a blog post about it like it’s some new found knowledge”.
The most annoying thing about Hasan or any other streamer on any point in the political axis is their constant need to have “the right opinion” instead of bringing about any sort of actual change or reduction of suffering. They sit in a room, have all their meals delivered, and spout “the right opinion” to whoever their target audience is. Honestly, I don’t know how they don’t get bored of it. Probably all the money and attention helps, and some fake rationalization on how they’re the voice of reason and light in the world.
This is phrenology nonsense and it’s shocking to see people almost nodding along in the comments. This is the same kind of nonsense people spout when they say they’re great interviewers and “just know”, when actual studies show they very much do not.
The goal to automate is to reduce suffering. Full stop. It’s not to “save time”. STEM types like to pretend they’re stoic cold calculating robots and everything is objective and they don’t mind doing some repetitive 5 minute task every day, because they saw some xkcd comic about efficiency. Maybe they pretend they don’t mind simply so they can smugly post the xkcd comic every time someone new asks why they’re suffering through some repetitive slog.
All the comments reacting with hate because they know, deep down, it’s true. They’re not the main character. Nobody cares about their hyper encrypted nix home server with the perfect firewall setup. And they’re certainly not getting those hours of their life back.
It depends. If someone complains about something and then fixes it, I’ll take them over the toxic optimist. There is nothing inherently wrong with complaining, what matters is the intent behind it. A complaint can be constructive, mean, funny, or any other number of good or bad things.
Hahaha they’re cooked. GPT 4.5 was a massive flop. GPT 4.1 is barely an improvement after over a year. Now they’re grasping at straws. Anyone actually in this field who wasn’t a grifter knew improvements are sigmoidal.