Everyone is a coach these days. Id think better path is training camps with sanders, were athletes compete with or get paced by him
2 Likes
And TrainerRoad offers an A.I. triathlon plan for $20.00 per month. Iāve been using it since August.
Edit to add, I used it for several years starting in 2016 so it has a lot of data for me on top of the imports.
My experiences with the AI programs is they are helpful at scheduling things out and 80% of the time works, 15% of the time there are some not ideally placed sessions or even intervals within the sessions and 5% of the time to there is something entirely nonsensical like a single bike ride that day that is referred to as a brick session and makes reference to a run that doesnāt exist on the schedule.
I think the rationale applies that any consistency in workouts is going to yield big gains over time.
They might not ideal for taking a great athlete and making them into a world champion, but they are good for getting someone who has some reasonable degree of common sense autonomy to get moving and consistently do a program.
The top tier athletes or those with particular individual concerns would really benefit from a coach still.
I think that for 80% of the coached population AI can work, but only if they use their head and adjust as needed and are comfortable motivating themselves without needing a human demanding accountability of them.
3 Likes
I typically hold to the bike workouts, add more to the swim and definitely add more to the run/s. Itās a great āadvisorā based off analyzed data.
AI is great at processing large amounts of data but itās very far off from thinking for itself yet. It works with pre programmed info and historical data. In the context of training programs you still need a human to tell it what to do with the data, you need a human to write the base workouts etc and you need a human to give it boundaries on how to implement any changes and so on. It canāt interpret any nuances in communication and training feedback on its own and it canāt come up with something completely new/different on its own that isnāt already there in some form.
Iām doubtful of the usefulness of the conclusions it comes to regarding micro advice for training, injuries etc as it will be generic to a large degree still. I think itās most useful as a data crunching tool as of now and I donāt believe that part will change soon as we are not close to the next level of AI yet. That being said I can see AI training plans working well as a middle ground between normal pre programmed plans and an individual coach.
3 Likes
Yes - we are not there YET (and we may never see the end of this development). But within maybe 3-5 years, 95% of the triathlon market and probably 99%+ of the total Endurance/Fitness market will be better off using cheaper and smarter training programs developed by AI, rather than a human coach.
This shit is moving FAST.
We are currently working on deploying AI in the global construction industry.
IMO a place where AI (or just higher data processing) can make waves is if itās combined with continuous or frequent inputs of body measurements. Heart rate, hrv, glucose, stuff we havenāt even thought of yet. Thereās already continuous monitors for plenty of measurements, itās trivial to have them update every hour or so to a training model that then uses them to build the next workouts.
Would be really cool if it could build on itself from there. See what response was achieved by individual or combinations of workouts, building a model not just of a standard athlete but a model of the actual individual.
But theyāll probably just use it to sell Couch to 5k programs.
1 Like
We obviously agree.
However we argue that domain expertise is an important distinguishing factor to guide the training program development.
Our latest coach platform developments posted here to show you where we are up to.
For anyone that wants to assist and jump on the AI wave with us, weāve now opened our second round of funding you can read on here.
This is both a blessing and a curse of AI.
Quality of data matters immensely, both in training, as well as for the individual user. Garbage in = garbage out.
And unfortunately, a gigantic portion of the puzzle that remains largely non-objectively defined, is how you āfeel overalā', which is obviously super subjective, but super important. Iāve had periods of overtraining where I was crushing every workout, setting shorter race PRs left and right, but knew something was āoffā as I didnāt feel good at all, and had that āfriedā feeling day in and day out, but was too inexperienced back then to back off from all those tempting PRs. Then comes the massive underperformance on A race day despite all that heroic training. Maybe AI will get better at interpreting subjective descriptions of it and take it into account, but I suspect itās already going to be quite a challenge for AI to just outperform well-designed stock programs (like 80/20 plans) that are just incrementally a step up from your current level.
ChatGPT already can generate really generic, basic plans that have nothing wrong with them, even for a 24 week ironman build. Iāve seen them, and Iād absolutely have solid results with them, with no editing required. Itās that extra small percent of performance that you want to squeeze out of training thatās the challenge, as itās quite individual.
Consistent training week in week out also generates ok results without āAIā
These models are text predictors trained on a load of data. Thereās no intelligence per se.
What you get is something generic based on the training data that theyāve been fed.
It looks so āgoodā because there are a huge amount of plans out there that are more or less all the same structure. SBR each week. Long ride
/ run weekend etc. That makes the predictions good.
Added to that these things are trained and tuned with a large amount of human correction then have output templates designed to look intelligent - written in a way to make you thing wow, and, crucially, full confidence.
Itās a sophisticated Google search, which is why they will eventually pivot to search (which is what OpenAI are doing). At some point youāll get references in the answers and these will be paid for.
Itās a huge bubble.
For time being, though, people will see āAIā and lap it up.
Fwiw I run a software company and itās great at the moment. Anything we produce that does any kind of basic calculations is called āAIā powered, customers love it.
Itās bonkers.
Having said that, if you use it, do well then of course thatās 
.
2 Likes
What do many coaches do though (Im speaking from extreme ignorance here)-Im assuming hey have a basic library of swim/bike run sessions that they use to build up the calendar-they wouldnāt be tailoring every session specifically and uniquely to each athlete, just using the library with adjustments based on FTP etc?
They adjust for reports of fatigue or missed sessions etc, with adjustments for periodised training leading up to a structured taperā¦all these things should be able to be easily performed by āAIā. Essentially when you get it to factor in metrics like HRV, HR etc the āAIā should be able to perhaps put together a more effective plan than a human?
Same goes for Doctors, all they are doing is taking what they can remember from a text book and applying it to the patient in front of them. Throw in some human error and mental fatigue and a computer should be able to do a better, faster and more accurate job?
For the most part,coaches provide accountability and person to person motivation. Most people are more likely to hit training targets if they have somebody to be accountable to. With AI,you donāt get that and it would be much easier to let your consistency slide.
There is so much training,nutrition,excercise physiology and tech information online that people who are motivated can do their own research and coach themselves (It isnāt difficult). Coaches break all that down and implement it for those who want someone else to do the heavy lifting in programming.
Tri Clubs and training squads are even better for motivation to get out there and that is where AI,for me,fails. I am all about the grand adventure and training for fun and being social and rigid training programs just take that part of the sport away,hence why I have never had a coach.
People are tribal and want to be part of something which is why Lionel (to bring the topic back to him) will find a lot of coaching success with in person training camps.
3 Likes
One BIG factor of why having a coach keeps you motivated, is the ācommitmentā or coach/athlete agreement, that both parties promise to serve in the relationship.
I have worked (in a prior life) trying to automate this to big groups of athletes, such as a e.g. 200 corporate people in cycling, and it is not easy (to automate). The minute the personal aspect leaves the equation and it becomes to machine/programmed, people are falling off. A real person (in this case a coach) looking over your shoulder, checking in on your motivation, well-being, injuries, tiredness, is a HUGE factor for many people.
Can this be simulated in the future?? Yes - it can, and it will be.
1 Like
I see now, that there is a great new thread on AI vs Human Coaching. Should we move the discussion over there?
2 Likes
I hold a couple of certificates (Level 1 and 2) as a Tri coach, and I did some coaching for a few years (approx 40 athletes in total) until I got hit by severe impostor syndrome and decided that I wasnāt doing it anymore.
I think thatās about right, you build a massive library of sessions and a library of plans from those sessions, or at least thatās how I did. I had several base plans, build plans, race prep plans, etc etc. Those would be my starting point and then I would customize from there. Not everyone wants to SBR on the same days, or some people are out on a business trip and can only run, or their pool closed, etc.
But you are right, writing out individual sessions every time you need to prescribe a workout would be madness. And on top of that, if you are coaching age groupers, they have a full life outside of training, so itās important for them to know ahead of time what they will be doing. So Iād write out 2, 3 and sometimes 4 weeks in advance, and then simply adjust as they would give feedback.
I think the plan is the easiest part, especially since most age groupers will benefit from consistency, proper zone control, and good recovery. The hard part, or at least for me, was the human element. Helping them deal with their insecurities, doubts, questions, keeping an eye on their fatigue, putting their efforts in check since most of them will run harder than they should, but will swim slower, or take longer rest times than they should.
Writing the plan is easy, having them stick to the plan is the hard part.
2 Likes
So about that Lionel Sanders, what are your personal thoughts on what his coaching style will be?
Muscle confusion
Iāll be honest, I donāt know how many other coaches here are also athletes, but I was and itās two completely different universes. I am one type of person as an athlete, resembling very much other age groupers: doubts, insecurities, pushing too hard, worrying about ātoo much recoveryā, worried about missed sessions, horrible relationship with food, and the list goes on and on.
I did coach myself for one or two seasons, but I liked it better when someone was coaching me.
What I am trying to say is that Lionelās athlete persona is not necessarily who he will be as a coach. Itās easier to advise others, and advise yourself and follow it.
Iād be interested to see how the marketing of his coaching brand is vs how users actually experience it. I donāt see him moving away from the āNo Limitsā mantra in the marketing but then in practice I could see it not being as hardcore(for lack of a better word).
Lionel sez, āI can eat fifty eggsā¦ in a hour.ā