The World Anti-Doping Agency (WADA) has said it plans to harness artificial intelligence (AI) to fight illegal doping in sport.
WADA director Olivier Niggli confirmed that pilot projects are being planned for later this year.
In the struggle to be faster or stronger than their rivals, professional athletes have sometimes been known to cross a different kind of line. But from Lance Armstrong to the endemic doping that appears to dog Russian athletics, most of the individuals concerned are well-funded, highly motivated, and one step ahead of the authorities.
WADA, the global organisation tasked by sporting bodies to uncover drug cheats, is now looking to level the field with the help of AI and big data analytics.
VIDEO: Watch humanoid robots ski at 2018 Winter Olympics
Using AI to spot red flags
The anti-doping body has long complained that a lack of funding and human resources is inhibiting its mission to rid professional sports of drug cheats.
WADA has a huge amount of data at its disposal, from biological passports to athletes’ test results and activities. But modern doping methods are often so subtle that cheating is difficult to spot, even in the cold light of day.
Now WADA plans to use AI’s ability to spot the behaviour patterns that dopers leave in their wake and flag up suspicious events – information that human observers might miss.
In this way, investigations into specific athletes can be better targeted, rather than made at random or based on gut instinct, said WADA. AI will also allow the organisation to speed up its work by crunching higher volumes of data.
Read more: Pyeongchang Winter Olympics to be defended by drone-catching drones
WADA ya know?
“Only sophisticated algorithms would be able to spot the differences [in suspicious patterns of behaviour], which would allow the anti-doping organisations to focus on the right individuals,” Niggli said.
“Anti-doping organisations would potentially get a lot of intelligence by being able to analyse a lot of this data and immediately spot anomalies that may be signs of doping.”
So while doping techniques may be growing in sophistication, Niggli hopes to narrow the gap between authorities and cheats by looking beyond traditional methods.
“I hope that in five years we will be much better at analysing all this data that we have, and which are already collected,” Niggli said. “It’s a complex world which requires complex answers.”
Read more: British Athletics deploys digital pace-makers for Rio Olympics
Internet of Business says
News that WADA is on the track and limbering up for the starting pistol is encouraging, if this deployment of AI can help to make top-flight sports fairer, more transparent, and more honest. And it will certainly have a deterrent effect if doping athletes feel that they can no longer game the system: more good news.
But there are caveats.
Niggli’s comments may suggest that the organisation won’t just be looking at current data, or information gathered from recent sporting events, but also at historic data – which could date back to when these types of record were first digitised. If true, AI has the potential to rock the world of sport to its core – at every level – if doping behaviour is as widespread as some allege.
Most would argue that this would be a good thing. It would certainly be a good advertisement for the power of AI, but a knock-on effect could be to undermine the public’s faith in, and support for, all kinds of sport.
WADA should be cautious about publicising its future findings too quickly, as AI systems can exhibit biases, depending on the types of training data that have been added to the system, and on the parameters of the algorithm itself. Beliefs that certain countries, athletes, or types of sport are more prone to doping than others might be completely true, or they might be biases that could taint an AI system at source.
“Gut instinct”, to use Niggli’s phrase, might be acute professional intuition from years of experience in the field, or it might be simple bias.
In other words, WADA needs to be certain that its findings are correct rather than merely suspicious, to avoid an AI-driven media witch hunt.
And that means that there need to be other systems in place to verify that any athletes flagged by the AI system have actually used performance enhancing drugs or supplements; logic suggests that this would be just as difficult as before, as those athletes have evaded detection so far.
Any media hysteria about using AI to “find the guilty!” across every field of human endeavour wouldn’t necessarily enrich human society; it could debase both it and the technology.
In short: a good technology and a good idea, but handle with care.