How you and your business can start your AI journey

Panel: Judith Faherty (moderator), Stephen O’Dwyer (TrojanTrack), Diarmuid Byrne (EquiRatings) and Prof Tomás Ward (DCU)

“AI is powerful but not magical,” said Professor Tomás Ward during the final session of The Irish Field’s AI in Equine Conference held at Naas Racecourse on Tuesday, titled ‘How you and your business can start your AI journey’. Professor Ward was joined on the panel by Stephen O’Dwyer, founder and CEO of TrojanTrack, and Diarmuid Byrne, co-founder and managing director of EquiRatings.

The DCU Professor was somewhat of an outlier on a day when experts from a variety of equine fields spoke about how AI is transforming their business. “I don’t know anything about horses, but I do have broad, deep experience in AI and machine learning, applied to a large number of industries and a number of successful spin-outs using AI.

“I’ve seen where AI fails or when business is utilising AI when they fail and when they succeed. I’m positive on AI, but that doesn’t mean I’m some sort of zealot or evangelist for AI. You have to be sceptical, and you have to control it,” Professor Ward told the engaged audience.

With more than two decades of experience in human-centric machine learning research, Professor Ward gave four take-away messages for individuals and companies looking to start their journey in AI. The first was to believe the hype; followed by not undervaluing your human expertise and experience as that is your “competitive edge. AI is to augment to give you superpowers, but currently AI is too volatile and erratic to be fully trusted; it must always be used in the human loop.”

Thirdly, Professor Ward said you must focus on your return in investment - what problem are you trying to solve and will AI help improve the metric that is important to you from a business perspective. “People get so excited and jump feet first in - that is probably the biggest failure point.” Finally, he recommended starting small and experiment with AI. “Keep your eyes on the prize of your return on investment and, when you see something begins to work, experiment with scaling it up.”

Story-telling

Diarmuid Byrne is the co-founder and managing director of the sport horse data analytical company EquiRatings, founded 10 years ago with his best friend and Olympic event rider Sam Watson.

Byrne explained how EquiRatings - which predicts how the sport will go, as well as producing a body of media work - started out with a very basic predictive algorithm on Excel, which has been transformed and massively scaled up with the help of AI across all parts of the business. EquiRatings’ work also extends to risk management, which was one of the main pillars of the business in the early years.

“Human connection is really important and sport is one of, and will be one of, the protected areas when in an AI world where there’s so much certainty and accuracy over outcomes. I think the ability to tell a story and create connection is very human and will remain very human,” Byrne said.

EquiRatings built a fantasy game called Eventing Manager eight years ago, which cost between €100,000 and €120,000 and took close to 10 months to build, and was costing €15,000 a year to maintain. Byrne explained: “You don’t have betting as an engagement tool in eventing or jumping and gamification is a way to try and get fans to feel something. I got rid of the game, because it wasn’t making us any money and the tech was kind of dead. We’ve just built a new one [using AI] which took about four weeks and cost €60 or €70. And now we are good at it.”

Professor Ward added: “The key word there is that you’re good at it, because you’d built it before, you knew what you were doing.”

Data overload

Stephen O’Dwyer, founder and CEO of TrojanTrack, an Irish equine technology company pioneering smartphone-based AI gait analysis, spoke about the danger of data overload in a ever-growing data-driven industry.

“There is a fine balance between whether you need a college degree to read the graphs. So, if a trainer has 50 horses and we’re pumping out eight A4 pages per horse of data graphs, they’re just not going to use it. They’re not going to read through it. They could miss things that, if you had only read one report, you might have caught.”

TrojanTrack has broken the data down into traffic light systems - red, amber, green - followed by reports for the trainers and owners. When one show jumping rider asked what was the recommended rehab for the horse off the back of the report, it clicked with the company that that is where the real value will be - if they can provide a solution for any lameness.

Using feedback from the reports, TrojanTrack went deeper into research around recommending rehab off the back of their analysis. “That is where users can find the real value in the product,” O’Dwyer added.

The AI in Equine Conference was supported by the Department of Agriculture, Food and the Marine under the ETS Scheme.