The Olympics are happening, and they are monopolising the attention of sports fans all over the world. We are no different, and we missed them last year. One of the worse aspects of the pandemic was how it took away our ability to watch these massive events, so the return of the EUROs and the Olympics has been very welcome.
We haven’t been idle this last year, however, and we spent a whole lot of it developing the first working version of our MVP.
We have developed AI that can understand what it’s seeing on video, and we can train it to evaluate it. Our first working version is focused on football freestyle tricks. It sees the video, it knows what it’s seeing, and it can faultlessly evaluate it according to whatever criteria we provide it.
It has been doing it successfully for a while.
So, why not in the Olympics?
The future of refereeing is digital
Not all of it, of course, and there will always be some elements of sports that will need a human being to apply human criteria and judgment, but a lot of refereeing is technical.
And AI can do that better than the human eye.
The Tokyo 2020 page lists almost one thousand human officials at the 2020 games this year, which means that there’s almost one official for every ten athletes. It’s a brutal amount of manpower that is being deployed sometimes with the simple mission of measuring something that a machine can measure better.
Here are some examples:
AI can judge gymnastics and figure skating better than people
We invite you now to take a look at how gymnastics are scored.
The scores are calculated through a mathematical formula. That formula has two components, a Difficulty Score and an Execution Score. Once the judges determine the elements of each one of these (and only the highest ranking judges are allowed to do so), it is up to the remaining judges to stare intently at every individual gymnast’s performance and try to objectively declare whether things such as the angles of every limb are correct, whether the grip of both hands was released at the same time, whether the gymnast left the ground after dismounting, etc. etc. in a succession of minute details that goes on and on.
Let us tell you something right now: an AI will do a much better job of objectively evaluating these things.
And they’ll do a quicker job too.
Look at this video:
https://www.youtube.com/watch?v=fuzSbSU3cqE
This is what happens between the end of a competition and you actually knowing who won. The teams just stand there, holding hands, in terrified silence, waiting for human officials to pick apart their performance.
Mind you, this is not the officials’ fault! They’re the very best people for the job!
They’re just not machines.
AI can judge any race better than the tech we’re using now
The tech we are using to evaluate races (this includes both athletics and swimming) is pretty impressive in its own right. It’s essentially a super evolved version of the same photo finish equipment that’s been used since the 1960s, and it captures over 3000 photos per second to track who wins.
But who wins is just the start of the conversation.
By using AI to track the movement of every single athlete on the pitch, we’d be able to instantly know not only who wins, or who commits a false start, or who commits any kind of foul during the race, but we would accumulate tons of significative information, such as:
- The relative speed of every runner at any moment;
- The way athletes are using their bodies to achieve peak performance;
- Where mistakes lie;
We could improve refereeing, and make it more than it already is. Aside from assessing rule-breaking, the very same mechanism could assess performance and actually help athletes become better at what they do.
This is the future
Now, these examples are real.
We are working on making these things happen. We are building these solutions. We have already seen what the first iteration of our technology can do, and there is no way a human person could match it.
Are you ready to join us in building this future?
