Posted on 2020-12-17 by: Axel Kloth
We as a group have discussed and will go into much more detail on how AI can be biased due to humans feeding AI “bad” (i.e. biased) data, and how that can be avoided. My take on it is very simple, and I have voiced it and it’s published. If training an AI system, make sure you let it have all data – the good, the bad, and the ugly. Here is a link to my fireside chat. Axiado is pointing to the same chat. I hope that the chat leads to a discussion on AI, ethics and our responsibility as scientists, engineers, users of AI, and of course as citizens. I don’t claim to have the penultimate insight into it, and I have failed to find a way to instill ethics into humans in a reliable and irreversible fashion, but I think that this is good enough as a starting point.