Artificial Intelligence is clearly the biggest story in tech right now. The success of OpenAI’s products like ChatGPT and DALLE-2 has taken the world by storm. The former, for example, now has over 100 million users, despite only launching in November 2022, which makes it one of the fastest-growing tech products in history. Experts are now saying that the disruptive power of these tools will be on a level with the arrival of the world wide web in the 1990s, transforming business, society, economics and culture.
Of course, AI has been around for a very long time. It was back in 1997 when IBM’s Deep Blue beat Grandmaster Garry Kasparov in a chess tournament, which will forever remain a milestone in heralding the age of the supercomputer. It’s also over a decade since IBM’s WATSON became a Jeopardy champion. But experts believe we are seeing an acceleration in AI and machine learning, and that will really start to take hold in the coming months and years.
Another area where you see AI being used is in sports predictions. Indeed, before a major tournament like the World Cup, you will always see tabloid-friendly articles covering what a supercomputer has predicted. However, more often than not, these predictions are badly off. To give you an example, an AI made predictions for the (ongoing) Premier League 2022/23 season before it kicked off last summer. And the majority of its predictions were either predictable, or plain wrong.
AI Missed Arsenal’s Ascendancy
For instance, Arsenal are the clear favorites to win the title in the Premier League odds at Mr Green online betting, as well as every other major sportsbook. It looks likely that the Gunners will win the Premier League, or finish second at worst. The AI predicted a fifth-place finish for Arsenal. But it gets worse: It predicted that Liverpool and Chelsea, both of whom have struggled and remain in mid-table, would finish second and third respectively. Of course, there’s always the chance that Arsenal have a dramatic collapse, and Chelsea and Liverpool have a massive revival, but it looks hugely unlikely.
One of the “problems” that AI has is that it has difficulty dealing with something called unstructured data. You might see unstructured data as somewhat – but not wholly so – synonymous with human intuition. For example, Chelsea have been rocked by the sale of the club, managerial changes, and influxes and outfluxes of players. A lot of football fans could sense instability. But on paper, Chelsea have had good results, spent money to bring in good players, and have an astute manager in Graham Potter. Computers generally favor the latter view.
Unstructured data modeling will improve
Where AI has the advantage is in analyzing structured data, i.e., results, form, impacts of specific players, injuries, and so on. We might be able to read a few expert form guides, but a computer can process millions of data points in seconds. Perhaps Barcelona lose more often when it’s raining, or Novak Djokovic struggles in the subsequent tournament after playing five-set tennis matches – those small data points, and many others, can be instantly picked up and processed by an AI.
The key, perhaps, is that experts maintain that AI is improving all the time with unstructured data, so we may soon see a tipping point where it becomes vastly superior in sports predictions compared with the human mind. Will that day be soon? Quite possibly. We opened this piece by talking about acceleration of the technology, and many believe that we are already at the tipping point.
Of course, not even the most powerful computer of the future will be able to be 100% correct with sports predictions. It’s part of the beauty of sports that the outcome cannot be predicted in its entirety. An AI might get close to perfect, but the competitiveness of sports means there will always be some mystery to the final result, or else it wouldn’t be sports.