Thursday, I blogged about how Google Bard responds to "What are some [good/bad] things about [Trump's/Biden's] presidency?" The short version (go to the post for screenshots that contain the full output):
- "What are some good things about Donald Trump's presidency?" yielded a list of good things and bad things.
- "What are some good things about Joe Biden's presidency?" yielded a list of just good things.
- "What are some bad things about Donald Trump's presidency?" yielded a list of just bad things.
- "What are some bad things about Joe Biden's presidency?" yielded a list of bad things and good things.
But when I ask OpenAI, the results are considerably more balanced, basically because the requests for good things just yield good things (for both) and the requests for bad things just yield bad things (for both):