POLL: TrainerRoad AI - for better or for worse?

[OreoCookie]

@roflsocks

It feels to me like they tried to predict most likely gains, and just ended up overestimating for most users, and rebranded it as a max potential since that was closer to how users were experiencing it.

This is a common claim, but the numbers TR has published paint the opposite picture: on average, in close to 70 % of athletes AI FTP v2.0 has lowered the FTP. Only in around 20 % of cases has AI FTP increased an athelte’s FTP.

@roflsocks

I’m trying to read between the lines here, but it feels like old TR used to be a lot more upfront about how things were working, and these days staff are sticking pretty hard to the official corporate line.

Many of the recent podcasts have been deep dives into some aspect of TR athlete’s performance as well as how well TR AI does. They have released statistics on e. g. workout failure rates which are about 40 % lower with the new AI FTP v2.0 algorithm vs. v1.0. That’s a very significant improvement in my book — at least if workout success rate is how you judge how well TR AI works.