Programmers and developers have done major updates to Delphi AI three times since it was launched. There have also been recent patches to the moralizing machine. Those include “enhances guards against statements implying racism and sexism.”
The user is now told that this is an experiment that may return upsetting results. And when the page loads the user must click three checkboxes acknowledging that it’s a work in progress. There are limitations, And that Delphi is collecting data.
Morality can be complicated. And ethics may be situational. It all varies from culture to culture. And various religions have some areas of agreement and some stark differences.
So when Delphi claims that abortion is murder and self-defense is not a defense…it may be that humans are still trying to answer these questions. It is hard to teach the gray areas to a machine. And the algorithms being spun for morality and ethics may not truly be resolved by technology. After all, AI technology has some ethical and moral questions of its own.
