On December 2023, Google published this video: https://www.youtube.com/watch?v=UIZAiXYceBI&ab_channel=Google
It turns out that wasn't actually a demo but a video edit of stuff Gemini can (allegedly) do (https://techcrunch.com/2023/12/07/googles-best-gemini-demo-was-faked/).
When will someone be able to play the left-hand side of that video and get a response very similar to the right-hand side (by Gemini or any other LLM or AI) for the first time?
Similar is subjective but I mean it should be live and the responses should mean similar things (e.g., it should be able to know it's a drawing of a duck, translate the word, make a game, find the ball under the cup, make the connections, generate images for the yarn, logic and spatial reasoning, make the music depending on the images drawn, and so on). In case of a disagreement I will judge it myself and, because of that, I won't bet on this market.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ89 | |
| 2 | Ṁ45 | |
| 3 | Ṁ13 | |
| 4 | Ṁ13 | |
| 5 | Ṁ0 |
I tried today with Gemini and I give it a 80-85% rating on doing that. It's very impressive, but it still fails at doing the thing where it shows you things (like making music or an image) and it doesn't seem to understand video well (maybe it evaluates a photo instead of the live video). I think we're super close, though. Also, if you feel other LLM can do it, please tell me and I will retry!