Has anyone tried any AI coach platforms? Pillar, Tridot, Humango, there seems to be tons popping up!
I believe that @pav may have had something to do with one of those
Do we know what level the AI is at - like is it actually AI or is it just complex algorithms?
The lines are blurry. A few years ago, a young whippersnapper colleague sold a rolling average as “the AI algorithm” to my boss.
I’ve not tried those, but am aware of them and the recent advances in things like Chat GPT. Do we think this is going to put coaches out of a job?
Interesting. Will be checking some of these out.
Correct! I have worked on Pillar and I now work with Humango. Very interesting and certainly the future of cycling training/coaching - without actually replacing the coach.
Pav, that opens up an interesting set of questions. What does a human coach provide over AI (which at the end of the day is more or less advanced pattern matching) ? And then, what are the limitations of AI in coaching as of today ?
I am a tad biased there for once doing my masters in a research group with neural networks (the origins of all of that), as well as working for a company who does a lot with AI (i.e. NVIDIA). However having seen what TrainerRoad propagates as AI is more or less clever marketing with nothing real behind it.
I will give an example of what i don’t think an AI system will do. I was planning for a big event and went through a couple of psychological exercises around motivation. There was a questionnaire with a part for the coach and a part for the athlete. I read Pav‘s replies about how he knows my level of motivation and I blushed. Can an AI system deal with the non mathematical elements of building and adapting a plan?
AI will automate some of the jobs that take a while - data analysis for example. Or aid with quick adaptations, adjusting a session based on fatigue or time availability (for example, you get stuck in traffic and have to do something shorter, AI might respond quicker than a coach (although I feel I am responsive!!)).
Julian summarises it really well. Another way to look at it is that you might Google your symptoms when sick, but will go to a doctor for an actual diagnosis or treatment.
Essentially, what we will see is a change from anyone using static and prewritten (generic) training plans to AI. Coaching prices may come down (as the coach does less grunt work).
I would never feel a bit of liability towards an AI app. This is no replacement for a Coach for me. It could help to make a decission when you are not recovered well and it might good to knock it off a bit but you should learn to do that by feel. (I’m still stuggling with that, too much engineer in me).
I agree with this 100%
Such a fascinating discussion. I look forward to watching this.
Agree with this. I think the main challenge is bias of the athlete/user combined with AI making decisions based on rules or available information. Maybe in the very far future, AI will think like a human brain can, but looking at the way ChatGPT can have ‘hallucinations’, we might be a way off that, despite how impressive it is!