Dentsu, the Tokyo-based parent of Dentsu Aegis Network, Wednesday announced the deployment of an artificial intelligence program capable of predicting the outcome of sports competitions.
The agency holding company said the program, dubbed “AI11,” was used to predict the outcome of a soccer game in real-time utilizing actual match footage.
The company described it as “an innovative and fund spectator experience,” but the implications of accurately predicting the outcomes of live sporting events could have unforeseen consequences for ratings, as well as betting on events whose outcomes are known before they happen.
The program was first introduced as a pilot test at the EAFF E-1 Football Championship 2019, an international soccer tournament held Dec. 10-18 in Busan, South Korea.
advertisement
advertisement
“This program was developed through deep learning, for which approximately 480 past matches including the ones from the EAFF E-1 Championship were used to predict match results, analyzing elements such as ball and player movements, combined with Team Twelve's match analysis method,” Dentsu said in a statement, adding, “Real-time match prediction was provided during the 2019 E-1 tournament, and this viewing experience was offered on TV and online.”
Big deal. I can make accurate game predictions too. The Giants are going to lose.
Joe, there have been many predicting models before about all kinds of subjects. In fact I developed one to predict primetime TV ratings for new shows many years ago, using all sorts of data---set usage, the strength of lead-in programs and competing entries, age/ sex demos, etc. plus judgements about the assumed quality of each new series. We batted about 65-70% for the fall and went up to 90% in predicting the secood season ratings for January-February. The latter was no big deal as we already had a very good fix on how all but the new second season shows had done in the fall. I wish Dentsu luck---but I don't think that this will work much better for sports. Of course, I might be wrong---let's see.
Several years ago I predicted that someone with way too much time on their hands would develop game result predictors in real-time.
Looks like I was spot on.
P.S. I have a sports predictor that is 100% accurate. When there is less than a second left to play and the margin exceeds a single score, then the team leading at that time will win! Nifty eh!
The real question is did the AI program predict the next story down about Dentsu cutting three percent of thei workforce and downgrading revenue and profit outlooks?