I was going to respond on a complaint thread but decided it would just be lost. I think the new update is doing well overall. However, I find I’m very emotionally affected by the implied feedback.
I found this change in behavior to be challenging in the new AI update. The workout change approval was the most clear feedback on my training performance. Change to harder workouts meant I trained and recovered well.
I find a couple new feedback that replace it. First, each training zone has a score at the end of a workout. I can see for example, if I took a pee break during an interval, that zone will have a low score. The other indicator is the FTP prediction will go up or down based on my workout quality and recovery.
I would like to see more direct feedback. Something that says your workout and recovery was good enough we’re going to push you harder, vs. you are inconsistent or have poor workout performance, your training will be decreased.
For me, the other feedback I was used to was yellow and red days. I’m not getting way more red and yellow days which feels like scolding for poor compliance. I can’t tell if it is a side effect of showing the latest prediction for the the week? Or, if it is telling me I’m doing poorly.
So what you’re asking for I think moves more into LLM territory. Though an LLM trained on the Trainerroad data set and training model. They’ve said on the podcast this is something they’d like in future. I.e. The current AI (which is machine learning not LLM) just prescribes you workouts and then changes them based on what you’re doing but doesn’t tell you why. Just have to trust it. LLM would be able to explain why the plan has changed, give you feedback on rides, etc. Hopefully more meaningfully than the Strava AI…
2 Likes
Assuming this was now not not? I do the same all the time.
If you do in fact mean now getting more, that isn’t about compliance or performance necessarily it is there warning you. And in this case unlike other adaptations it is giving you the feedback you want. You are overdoing it and should back off a little. It may be that the system doesn’t have enough information about your history, so if that is the case you’d need to judge that.
In regards to the system telling you what its doing, it does. That partial credit is telling you that if a workout later changes and it changes down it is becuase you didn’t hit the target. This isn’t bad it is just changing it to a workout its pretty sure you can pull off instead of pushing you further and causing failures. You can use the predictors to see whats changing, you can observe if the workouts in the future are higher WL or lower ones to see what it is doing. This isn’t being spit out in spoken language like an LLM but it is there for you to read and understand. I can understand wanting it to tell you the why of what it did but you can also learn quite a bit more about it by observing what its doing.
1 Like
Agreed, for giving descriptive feedback, it would be LLM territory. I would love to build that capability.