What exactly does the AI do?

[gtech]

Does it collect data from all your tasks to deliver the insights? Does it also modify the tasks to train your weaknesses?

[Chester B.]

Originally posted by gtech:

Does it collect data from all your tasks to deliver the insights? Does it also modify the tasks to train your weaknesses?

Hi there! Thank you for trying Aim Lab and we're glad to provide support for you.

Basically, the AI takes into account last 100 plays and adapts your performance. For example, you are doing Rocketshot Long-term AI. If you do well in the far upper left of the screen with precision but not so well with speed, the AI will give you more targets in that area that start small but slightly out of your comfort zone on speed.

Let us know if you have any other questions. We will be glad to hear from you!

[gtech]

Wow that is SO COOL. The AI is meant to induce flow by going slightly out of your comfort zone?

I notice that it prioritizes precision over speed. Is there data behind that being ideal? My precision is in the 90th percentile but my speed is in the 56th. There must be some point at which you must favor one over the other.

The extent to which any one task or skill performance influenced others is also interesting. At this point I've been trying to give proportional time to each task with respect to relative percentile performance. Perhaps the AI could recommend one task over another in order to maximize gains over time?

I'm not sure whether aim-lab training should attempt to maximize k/d or matches won, there are trade-offs between difficulty of obtaining data, what kind of play-style you want to optimize for, and potential for confounding variables. That would require a survey of aim-lab skill performance vs outside game performance. How much are your choices of the current tasks and skills data-driven?

[JoshP]

Originally posted by gtech:

Wow that is SO COOL. The AI is meant to induce flow by going slightly out of your comfort zone?

I notice that it prioritizes precision over speed. Is there data behind that being ideal? My precision is in the 90th percentile but my speed is in the 56th. There must be some point at which you must favor one over the other.

The extent to which any one task or skill performance influenced others is also interesting. At this point I've been trying to give proportional time to each task with respect to relative percentile performance. Perhaps the AI could recommend one task over another in order to maximize gains over time?

I'm not sure whether aim-lab training should attempt to maximize k/d or matches won, there are trade-offs between difficulty of obtaining data, what kind of play-style you want to optimize for, and potential for confounding variables. That would require a survey of aim-lab skill performance vs outside game performance. How much are your choices of the current tasks and skills data-driven?

Hey gtech my name is Josh and I mostly deal with QA so I don't personally have an answer to these questions but didn't want to leave ya hanging. Someone with the proper knowledge on your questions will get to it, but if you'd like to join us on Discord we have a science channel dedicated to all kinds of questions/discussions such as these and other channels to discuss/suggest features/bugs along with a wonderful community of which you can join here: https://discord.gg/DamW5c5

Again, if Discords not your thing thats fine to, someone will get to your question asap, but just figured I would throw that option out to you as well.