https://www.newscientist.com/article/2406746-google-says-its-gemini-ai-outperforms-both-gpt-4-and-expert-humans/ Work is changing and it’s happening fast.
How do they measure the performance?
They reference MMLU in the article. https://paperswithcode.com/dataset/mmlu
https://paperswithcode.com/sota/multi-task-language-understanding-on-mmlu
Google can say anything they want, but until we see the final result, I’m sticking with OpenAI and their agility when it comes to winning the AI race. With how medicore Bard was I don’t expect Gemini AI to be anything more than a slight improvement over GPT4 at best, unless they managed to fix and remedy many of the existing issues with LLMs. Even then, I don’t trust the slow moving nature of Google to outcompete the agility and scrappiness of OpenAI, especially as a startup. It’s their one key strength when it comes to hiring superior talent. TC: 105K
Says a consultant.
And? I would have the same opinion if I was in any other position. OpenAI has superior talent and a more stable base to work from, unlike Google which is playing catch-up.
See it to believe it
The pro model is live today the ultra model comes out early next year.
Do you find it better than chatgpt?
How can I access it?
Why wouldn't Google say that? They've just slapped lipstick on a bard
Watch the video. There’s some seriously impressive shit
It's just that. A video. How many times will you fall for it
Aaaand there it is: https://techcrunch.com/2023/12/07/googles-best-gemini-demo-was-faked/
Just wait till GPT5 outperforms Gemini and then the other way around again.
Where is BlindGPT coming out
There’s one for levels.fyi so I agree with your question!