There is already a lot of research on predicting [1] and monitoring [2] wildfires using traditional machine-learning approaches.
How would an LLM be useful in predicting wildfires? A project that incorrectly claims to solve a problem "using AI" has negative value.
By proposing a solution that will not actually work, these hackathon participants make a complex challenge seem trivial and take away attention from other, more mature work.
EDIT: I checked out the project's description. It seems that they do not actually use LLMs but old trusty XGBoost.
It looks like they actually have put some thought into this - however, I'm still not convinced that it would work in practice at any reasonable scale. (Maybe some billionaires might want to invest into something like this to protect their own mansions.)
If a break through product comes out of a hackathon, that would be amazing. As far as I can see it after participating in a few of them, I think a hackathon is more than that. There social aspect of it is important, including the motivation and the inspiration of the community.
How would an LLM be useful in predicting wildfires? A project that incorrectly claims to solve a problem "using AI" has negative value. By proposing a solution that will not actually work, these hackathon participants make a complex challenge seem trivial and take away attention from other, more mature work.
[1] https://www.nature.com/articles/s41598-025-92171-w [2] https://sites.research.google/gr/wildfires/
EDIT: I checked out the project's description. It seems that they do not actually use LLMs but old trusty XGBoost. It looks like they actually have put some thought into this - however, I'm still not convinced that it would work in practice at any reasonable scale. (Maybe some billionaires might want to invest into something like this to protect their own mansions.)
LinkedIn post: https://www.linkedin.com/posts/emmanuel-karibiye-509b9a1b3_h...
Project site: https://zerostrike.live/
Description on Devpost: https://devpost.com/software/zerostrike