Can a Computer Rebuild Manchester United?
Humans have failed, over and over again, to create a competitive soccer team with all of Manchester United’s money. What if we let a computer give it a whirl?
I appointed AI as Manchester United’s director of football and asked it what the club would do this offseason:
OK, let's get into some specifics. Who's on your five-man shortlist for central defensive midfielder?
[Editor's note: ChatGPT suggested a number of players like Declan Rice, who are not available or way too expensive to acquire or both. After some back and forth, it landed here.]
My apologies for the oversight. A realistic candidate could be Sander Berge (Sheffield United/Napoli). Berge had shown promise as a defensive midfielder with Sheffield United before his move to Napoli. He possesses good defensive instincts, physical presence and passing ability, making him a well-rounded midfielder who could strengthen Manchester United's midfield.
Berge's experience in the Premier League and potential for further development could make him a suitable target for the central defensive midfield position within the specified budget range. Given Sander Berge's age, Premier League experience and potential, his transfer fee could range anywhere from £15m to £25m, depending on the negotiations between clubs and market conditions.
Congratulations, Pacers fans. Happy Tuesday, everyone.
Replacing Diogo Dalot - the player of the season and one of the only players who showed real promise and quality throughout the year - with max Aarons - and spending a net $20mil to do it. I think we've got a long way to go.
You should try Claude for this kind of experiment in future! LLMs are not great for this sort of thing because their training data typically cuts off significantly before the present, and Claude's is a year out od date so it did require a bit of help- but it goes about the project in a significantly more logical way than ChatGPT.
I've used it for similar experiments in American football, and I thought it produced a pretty cogent analysis, although I did have to feed it 2023 stats first.
Plus it's made by Anthropic, who actually care about AI safety (and in fact split off from OpenAI over disagreements on this). And it's way ahead of Gemini etc. from a performance standpoint.
ChatGPT and Claude are the leading LLMs at the moment, and Claude is the significantly more ethical choice of the two.