I’ve been running this experiment for a little over a week now and I can’t say I’m super surprised by the overall results: 19 for 40 – basically 50%
I’m keeping a running tab of each AI model’s record and they are all one pick within 50/50, no better than a coin flip. Here are few “far too early to tease any insights” stats that the models have shown so far:
- Copilot is 3 for 3 on NFL bets on 1 for 1 on NHL bets.
- Grok is 1 for 4 on NFL bets.
- ChatGPT is 3 for 4 on NFL bets.
- Perplexity is the only AI model to make a prediction for an non-U.S. league and it got it right at 1 for 1.
- My bets are 5 for 8 and I typically try to follow an AI consensus
Long story short, maybe I’ll give a tiny bit more consideration to Copilot’s NFL picks…maybe? I’m sure more information will be able to be teased out of the data at some point. I probably won’t dig too deeply into it again for another month, at which time I’ll have about 180 more data points.
Category: Analysis