We had Fry Daddy's for the first time a few weeks ago and I'm still dreaming about their fish and chips. Best I've had in a while.

But, here's an example of intentional manipulation of responses in grok. www.euronews.com/my-europe/20...

This is one of the biggest weaknesses with these LLMs. Asking them about current events gives really unreliable responses, because the limit of what they "think" of is now, is actually some point in the past.

While I completely agree with your concerns, and while it's possible that some intentional manipulation is being made to the responses from these models, it's also possible that some of these "that hasn't happened yet" responses are simply because the training data for the model is out dated.

I was in line at the post office and literally LOL'd at this.