Assuming they are trained on datasets that do not control for this classification, and are derived from the product of more than ten individuals I think it’s safe to expect, unironically, that yes the results would be more predictive. (ETA: but not grounded, and pointless as a survey.)
What about a focus group of 10 large language models?
Assuming they are trained on datasets that do not control for this classification, and are derived from the product of more than ten individuals I think it’s safe to expect, unironically, that yes the results would be more predictive. (ETA: but not grounded, and pointless as a survey.)