Optimizing Preferences by AI


optimising individual settings seems to be a big problem in Anki, as there are many parameters with unpredictable influences.

Why not use AI (artificial intelligence)?

A program that finds the optimal settings based on the previous data after entering the individual goals?

With kind regards

1 Like

It’s an interesting approach but maybe more like suggestions than hard-coded algorithms that suddenly changes the pace. That could even be done with statistical analysis, no need for AI.

AI is learning from statistics. The more data, the better AI.

Yes but you could go far with just plain code using statistics without getting involved with neural networks and such.


Where can I find advice on what to do when default values fail? Advice on the internet that I have found seems plausible at first glance, but it is not, because there are unpredictable side effects.
For me it is a typical black box problem that cannot be solved with logic.
Please give me a link how to handle the problem when default values fail.

To start with, looking at stats give you good short term and long term indications of the training levels and what could be good to change, especially the pace.

PS: We should avoid the ‘clippy’ AI if someone remembers the sad saga about that Microsoft thingie.

It’s a cool idea with lots of potential, but I’m afraid the development resources will be allocated on more pressing issues for a while.

Successfully integrating an AI (without causing harm) is not a trivial task by any means and looking at the 2.1 scheduler, which is still considered experimental, you can see that even slight changes regarding scheduling take a lot of time in the Anki ecosystem.

But hey I hope I’m wrong, 'cause it sounds awesome!

Yes baby steps. I would just be happy if there was a suggestion note based on stats showing what could be done for help with studying such as creating filtered decks with the problematic cards, re-scheduling due to weekends and so on.