dialtone wrote: ↑20 Jul 2024, 21:08
RacePaceDemon wrote:Zynerji wrote: ↑20 Jul 2024, 20:09
If it was run by a RAG ai, it would already be perfect. The weighting of what's actually race relevant and important is obviously NOT currently making the choices.
Rag ai? I'm not too informed with AI could you explain how it would work with race direction?
RAG is Retrieval Augmented Generation. Basically a search engine written by using the feature vectors generated by LLM.
It’s not clear to me how it would help in this case.
It's able to utilize open cv on all streams and after being trained on historic footage, it can detect important things (overtake/crash/etc) and switch cameras, prepare replays, and give conversational data to the presenters. It could also monitor time deltas and show <1s battles in popup windows. It could also pick out all external updates to the cars or monitor track limits.
I'm seriously surprised this isn't already being done. Especially the commentator chat. That could change the deliverable insight metric by enormous amounts. Especially when the RAG can pick out car failures that lead to a crash and such instantly, and the audience get the info quickly.
Edit: using RAG as the catch all term for this is insufficient now that I think about it. Its more of an Agent than a RAG.