Important point. Remember my experience in maintenance of systems, always considered it a thing to be considered up front. Also, as is pointed out, there are other reasons to forget. For example forgotten data cannot be misused. Which links into the increasingly important inclusion of security. Also mentioned, the need to still be able to explain a model, even after a previous model was updated or re-run for use with updated data.
Now That Machines Can Learn, Can They Unlearn?, By Wired, August 19, 2021
Companies of all kinds use machine learning to analyze people's desires, dislikes, or faces, but some researchers are now asking a different question: How can we make machines forget? A nascent area of computer science dubbed machine unlearning seeks to induce selective amnesia in artificial intelligence (AI) software, striving to remove all trace of a particular person or data point without affecting performance.
However, the notion of artificial amnesia requires some new ideas in computer science. Once trained, a machine-learning system is not easily altered or even understood. It will take virtuoso technical work before tech companies can actually implement machine unlearning as a way to offer people more control over the algorithmic fate of their data. Even then, it might not change much about the privacy risks of the AI age.
No comments:
Post a Comment