Ask HN: Is there a non-chat Turing test?

I've seen a lot of debate whether the current large language models (ChatGPT, Bing/Sydney, etc) exhibit sentience. Given the history of the Turing test, I think it's inevitable for these discussions to happen whenever there are advances in chatbots and NLP.

However, I don't recall any discussions about sentience a few months ago when image generating models were all the rage (DallE, Midjourney, Stable Diffusions, etc). There were plenty of discussions about whether the machine should be given artistic credit, which gets really close to the sentience discussion, but I don't recall it being explicitly discussed the same way we're seeing with the language models.

So this has me thinking -- how could you assess sentience for machine/models if it's been trained to do something other than chat?