A.I Coaching

[Theunissa]

So I've been wondering will Ai coaches replace real coaching. With them being cheaper and the way Ai is apparently evolving

[no calves]

Trainerroad is already there.

[SwissVan]

You cannot replace the human side of coaching

 

[Frosty]

Zwift has replaced real coaching too, based on how some people ride in a bunch.

[Paul Gleed]

11 hours ago, SwissVan said:

You cannot replace the human side of coaching

 

Agreed. How will AI let you vent when a session or a race/event doesnt go to plan and join in your jubilation when it does all come together. A good coach is someone that practices what they preach, been there and done it.  

[dave303e]

honestly there are so many 'covid coaches' out there ripping people new ones, those people are better off with AI. I have a good coach and will never replace him with AI, purely because I can have a conversation with him around things and we work together on that. We also have now got a good few years and many many PB's together so I trust him, it is a relationship. My previous coach did not last, we just did not gel and goals/targets were hard to align/communicate and this led to me not trusting what was on the plan and then not doing it well.

 

That being said though and as someone who has written production ML algorithms this does interest me. The Mrs had our baby in Feb, so she has been lookin at making a comeback into fitness. She is actually now on trainer road adaptive training. She did not want the pressure of having the week laid out for her to tick the boxes in case baby has a bad night and she wants the day off, timing, other stuff. She just wants to have some structure as and when she decides to ride. So she jumps on trainer road and hits train now and based off that it gives her a session to do. Honestly she has already started feeling better and she has not got off the trainer going that was too hard or too easy. So far so good. Watts are heading upwards and HR and weight is heading downwards. She has not got off the bike swearing that it is too hard or had her energy/breastfeeding negatively impacted by the training. She has also really enjoyed the slightly tougher sessions. So I have been quietly impressed, but this is not a high level of training/fitness, but it is a great start and a great way of getting back into it without forking out for a coach. For the average mid to back field rider, great option. That being said, she has followed a few of the zwift training plans to the T and also had good results so a basic generic plan at lower level is again not a bad option. It is just nowhere near as flexible.

 

Something that people always forget about AI. It is a bit like a child. The quality of the material you teach it from determines the quality of the output. So if it is looking at a million bad programs it will give you a bad program. If it is looking at a few good programs then it may work if those programs had similar goals. But each person has slightly different needs/wants. I think at a basic level any basic couch to 10k or similar laid out program works. But as the athletic performance goals increase- it takes more and more input and more and more skill to get a program aligned to stretch that elastic far enough in the right ways and not snap it.

There is a place for AI in training but not everywhere and for everyone.

[bleedToWin]

When it comes to problems that people are able to accurately characterize (or even parameterize) AI can do well. Even when these problems are complex with many variations, if there is enough content out there with someone saying okay in this instance because of these reasons the solution looks like this, then AI can train on it and even start filling in some blanks.
I am yet to see a video or read an article on athletic preparation that looks at the holistic, gives parameters and gives a solution. They all say "it depends". AI has plenty of good training sessions and content on training principals to learn from however.  

[KidCharlemagne]

"Don't toss the baby out with the bathwater" as the saying goes. There's a function for both, the human coach and the ai based option. That choice is largely individualistic. Some athletes need that human interaction, but a computer-driven program is probably just fine for the average middle-packer. The more Luddite-leaning coach may resist technological progress while the progressive ones will use ai to complement their coaching practice and thus be able to manage far more clients.

[shaper]

You get 3 types of athlete

1. Either wing it or download a plan of the internet and follow it.

2. Self coached

3, Has a coach for not only putting an individualized plan to achieve a goal, but also for human interaction, accountability and specific needs let alone any one-to-one coaching tat AI cannot provide.

At present AI will likely be taken up by the Type 1 athlete.  A Type 2 athlete may consider it and then look at the structure of the plan and adapt it further.

Type 3 will always stay with a coach unless circumstances change dramatically i.e financial/long term lay-off etc in which case they will likely take a break and return to a coach when needed.

[Frosty]

A coach is looking to make decisions, and adjust if needed, based the data seen as well as the subjective feedback from the athlete. 

Data science has been getting more advanced, and for coaches AI can help them save time by providing quick answers to complex data capturing processes. 

Posted too soon… here’s a podcast link where Alan Couzens discusses integration of AI into coaching development.

https://www.fasttalklabs.com/fast-talk/neural-networks-possibly-the-most-important-training-tech-youve-never-heard-of-with-alan-couzens/

Edited June 19, 2023 by Frosty

[Theunissa]

On 6/19/2023 at 9:23 AM, shaper said:

You get 3 types of athlete

1. Either wing it or download a plan of the internet and follow it.

2. Self coached

3, Has a coach for not only putting an individualized plan to achieve a goal, but also for human interaction, accountability and specific needs let alone any one-to-one coaching tat AI cannot provide.

At present AI will likely be taken up by the Type 1 athlete.  A Type 2 athlete may consider it and then look at the structure of the plan and adapt it further.

Type 3 will always stay with a coach unless circumstances change dramatically i.e financial/long term lay-off etc in which case they will likely take a break and return to a coach when needed.

I'm no 3. I Complain alot to coach. Especially of I had. A hard working week and still have to train. When I slack the real coach will say. If you didn't use it you lose it. He also looks at my training and we talk strategy before a race. Example was the amohela. I was given a certain plan to work with to finish well on second day. It also does bring the accountability part. I'm not gonna worry if I miss if a computer program is gonna check up. 

[Theunissa]

On 6/19/2023 at 9:29 AM, Frosty said:

A coach is looking to make decisions, and adjust if needed, based the data seen as well as the subjective feedback from the athlete. 

Data science has been getting more advanced, and for coaches AI can help them save time by providing quick answers to complex data capturing processes. 

Posted too soon… here’s a podcast link where Alan Couzens discusses integration of AI into coaching development.

https://www.fasttalklabs.com/fast-talk/neural-networks-possibly-the-most-important-training-tech-youve-never-heard-of-with-alan-couzens/

So it becomes an efficiency tool. Like me using a battery drill and a meggar to find a fault on a cable. It helps you focus on important stuff amd not stupid stuff. 

[shaper]

https://www.trainingpeaks.com/coach-blog/biopsychosocial-coaching/?utm_campaign=coach2023&utm_medium=email&utm_source=newsletter&utm_content=cta&utm_term=20230625_coach