The Prediction Paradox – VAR for AI
As some of you may know I suffer from an unfortunate and at time very upsetting social condition – I support Leeds United. Last season was some of those rare happy occasions when things went well, we had a bit of luck at the right time and played some decent football, resulting in winning promotion back to the English Premier League.
Apparently this is our rightful place, but as in business professional sport delivers its own verdict based on results rather than wishful thinking and we have only spent 3 of the last 20 or so years actually in our rightful place – the rest of the time has been spent languishing in the 2 divisions below the Premier League. Not unexpectedly Leeds United are one the bookmakers (another objective impartial group of individuals) favourites to be relegated straight back down to where we don’t apparently belong, The injustice of it all – makes you want to cry.
This prompted me to wonder what the betting odds were on Leeds United defying the bookmakers expectations and avoiding the drop at the end of the season. Last time I placed a bet of any sort I was actually at a race course so odds were easy to find – but not so the odds for this flutter. The bookies in the village I live in has long gone. So I resorted to Google.
The only thing this did not tell me was the odds I required, but I noticed in between the 3 or 4 Hundred free bets that from what I could see seemed to cost about twice as much as the sum offered was a META AI – the AI engine powered by Meta no less, a company which I think it fair to say has a lot resources at its disposal.
In my excitement to leave the cascading torrent of betladfred365paddywilliam888 I clicked on the link and typed the question what are in the odds of Leeds being relegated from the Premier League in the 2025 to 2026 season? Off it went zinging around the internet collecting data form every conceivable and presumably inconceivable corner – diving into the subject in way that a mere mortal pessimistic Leeds supporter (for there is no other kind) has no chance of emulating.
I marvelled at the speed that it came back to me with, ignoring the fact that Blackburn had been plunged into darkness for a split second due to power required for my query, here was the answer from a tool that depending you r view is going to or already has changed the world irrevocably.
In seconds flat it had compiled a list of all 20 teams ranking them in order according to the place they are expected to finish. Now the speed did not surprise me – it after all only 20 teams – all based in one defined category – the Football Premier League in England for the 2025 – 2026 season. Then take the wisdom of the internet and rank them in where AI expected them to finish. Disappointingly the Meta AI engine reached the same conclusion as every punter from the pub to the sports pages and pundits beyond – Leeds are destined for the drop.
However what did surprise me was that one of the leading AI platforms also predicted that Sheffield United were predicted to finish last and also be relegated. This was an amazing outcome and one that offers a cautionary tale to any person or organisation that relies on AI or plans to rely on AI without checks and balances, because Sheffield United are not actually in the Premier League having lost in the play offs to Sunderland. Don’t know what that means for Sunderland hopefully they will survive by virtue of Meta not knowing they are even in the Premier League. In terms of AI I would have expected a query based on ranking 20 entities in order according to would have been a walk in the park but it got it completely wrong so be careful when you ask the question and perhaps take time to check the answer…
Footnote – at the time of writing Sheffield United are bottom of the Championship so perhaps it is smarter than I think 😊
Phil Madders, Managing Director – PAE Business
