6 Comments
User's avatar
Samuel  Anderson's avatar

Go ask your local store, bar or restaurant. They will tell you what is happening when it comes to drinking culture. Rather than me telling you here, go into the wild and observe, ask and report back. Definitely, changes from generation to generation. By the way, love the article you shared. It was wonderful.

Expand full comment
Felicity Carter's avatar

Thank you very much!

Expand full comment
Samuel  Anderson's avatar

Keep sharing.

Expand full comment
Joe Fattorini's avatar

I enjoyed this round up very much. I read somewhere that a large amount of training data also came from Reddit. This led to the curious feature that LLM’s are prone to using Star Trek analogies and insights. Beam me up Scotty

Expand full comment
Felicity Carter's avatar

Thanks, Joe! I have tried to train ChatGPT out of using certain words ("fluff") but it can't be done. It also still indulges in "lived experience" though I have forbidden it. It just writes "lived experience (sorry)".

Expand full comment
Joe Fattorini's avatar

That’s utterly fascinating. The idea that the training data suggests the best way to deal with that query is to ignore it (and continue to predict ‘lived experience’ as the most likely words to resolve your query) and apologise, rather than accept the instruction and change behaviour.

Expand full comment