Also on their website: Saturday Morning Breakfast Cereal - LLM
Hover text: Best you can do with cosmos-3.5 is get a universe that’s classical at low speeds.
Bonus panel
RSS Feed: https://www.smbc-comics.com/comic/rss
This sums up in a funny way what I’ve been thinking for a long time: There is often disparaging talk about ChatGPT being “just” an LLM. But the whole logic behind it is not that different from many natural processes.
We refer to ChatGPT as “just an LLM” because there are weirdos that think it’s Skynet. ChatGPT isn’t going to take over the world. The cyber apocalypse isn’t just around the corner. It can barely string together 2 sentences without one of them being outright false. We are a long way away from it subjecting us, and even farther away from it simulating a realistic universe with artificial life thinking that they are building an interactive chat program that could one day simulate a realistic universe.
Although I agree with you, I have heard the argument in other contexts.
The “It’s just an LLM” card is quickly drawn as soon as minor errors or anomalies become apparent. It’s “just a machine that strings together somewhat logical sentences.”
Even though this ability makes the program more intelligent than some people I know.
