K-AI: Enable to provide feedback to improve K-AI feedback (Reinforced Learning)


when using K-AI the code preview showing delta is quite nice but lacks, from my point of view, three important aspects:

  1. Provide feedback to code changes from K-AI
  2. Rendered preview as the code sometimes does not work
  3. History to revert or start from an earlier query point (also using that as a reward / penalty)

Worth to note that this might lead to AI-Model-Poisoning but that is all up to the design of the process.

Most importantly, if K-AI can leverage the feedback by it’s users, it’s soon becoming all all sentient … joke aside but here is a meme :wink:


Hi Mike!

Good ideas and thank you for your meme :joy:.

About feedback: Yes, right now we are not collecting it. But that is certainly something that we want to have a look at in the future :slight_smile: