Can AI turn you into a pro athlete? Half-marathon training diary 2

Nathan Limm is putting AI to the test. Over 10 weeks, Microsoft Copilot is training him to take 15 minutes off his half-marathon time, using information on his physiology and daily training logs. Success could point to potential for a generation of top athletes to be coached by AI.

Four weeks in...

Patience

Like most 20-somethings, I lack a certain sense of “calm” that (I hope) comes with age. My issue when it comes to training is patience. When I’ve run my own training programmes in the past, I’ve been guilty of going too hard too early. I want to run long distances with intensity, now.

With an AI system at the helm, I have been eased into the training very gradually. In these first few weeks, I’ve been running frequently, but for only 20, 30 or 40 minutes, between five and seven kilometres. It’s been slightly frustrating. I’m not running anywhere near as hard as I thought for taking on a goal as ambitious as slicing 15 minutes off my half-marathon time. To put it simply, it’s been rather boring.

I have to give the AI credit. It has eased me out of a cold and managed a niggly lower-back issue while keeping me running five to six times per week. Nothing about this has been flashy. Simple mobility sessions to stay on top of my back issue and slow runs along flat terrain have dominated the schedule, with just one lower-body strength session so far.

One thing I have loved about the training is the adaptability. I’ve weaved in 40 minutes of social netball each week, giving the AI an approximation of the intensity and how the movement has affected me. It’s an acknowledgement that I’m not a professional athlete, and I have a normal life that provides disruption to what would ideally be a blank slate to fill with training.

Nathan Limm: "One thing I have loved about the training is the adaptability." Photo / Annaleise Shortland

Confusion

I have noticed a few hiccups in the fluidity of the conversation with my AI coach. Despite having access to the entire sum of all human knowledge, Copilot has been confused on a couple of different occasions about what day of the week it is. I’m unsure exactly why – given the chat even says the day of the week when you look back at it – but I have had to correct the AI to get clarification on what the training week looks like when, on a Monday night, it is referring to “tomorrow” as Wednesday.

Microsoft's Copilot got confused a couple of times about what day of the week it was.

Reassurance

As mentioned, the first three weeks of the training have been somewhat disrupted by illness and injury. It has meant I have missed out on the weekend’s proper long runs, designed to help me run further rather than faster. Given the relatively tight, 10-week timeframe, I voiced my concern to the AI about whether my goal of sub 1h 40m is still realistic.

Microsoft's Copilot has provided reassurance when I have been nervous about missing training sessions due to illness and injury.

Copilot gave me a lengthy, detailed explanation about why the weeks to come are far more crucial in developing speed, rather than the initial training block. It’s also begun weaving reassurance into other answers as well, lowering my anxiety about progression.

Admittedly, the AI has helped me feel better and encouraged me to persevere. It remains to be seen whether the results will justify the optimism.

Nathan Limm has been a journalist with Newstalk ZB and the NZ Herald since 2020. He covered the Netball World Cup in Cape Town in 2023, hosts The Big League Podcast and commentates rugby and netball for Gold Sport.