Some authors consider the not a good test for finding out if an is really “intelligent” because the test is "*purely behavioral*" and looks only at responses without investigating the mechanisms that produced them.

I think that asking for the "mechanisms" would in fact introduce unnecessary bias and invalidate the test. The test is solid in principle, it is just that we have to come up with some better test cases (scenarios).

Follow

@Kihbernetics

Use a modified Turing Test with specific questions designed to assess thought processes, like psychological examinations.

Then make sure the AI is never exposed to those questions or their antecedents during training.

Then when you ask those questions after training, you will be able assess whether the AI is actually sentient or not.

Sign in to participate in the conversation
Qoto Mastodon

QOTO: Question Others to Teach Ourselves
An inclusive, Academic Freedom, instance
All cultures welcome.
Hate speech and harassment strictly forbidden.