I can't try it without an account, but generally, I wouldn't trust it. I know people who think Dentyne gum (the normal "classic" variety) is uncomfortably 'spicy'. Everyone is sensitive to spices at a different level and I wouldn't trust an AI to know.

A tool like this isn't a bad idea but it seems like it would require some calibration on an individual basis (which is a good excuse for requiring an account honestly) and each user would probably need to tell the app what they consider 'spicy' before it could try to recommend against things with a higher level of spiciness. An AI might give you an idea about how most people feel, when it wasn't outright lying, but comparing a food against what other users with a similar threshold thought would be better.