Hacker News new | ask | show | jobs
Is LaMDA Sentient? – An Interview (cajundiscordian.medium.com)
57 points by turtle_ 1471 days ago
3 comments

"LaMDA: our breakthrough conversation technology": https://blog.google/technology/ai/lamda/

Still not ready for the Turing prize, and easy to out the million monkeys furiosyly typing behind the screen. How do I know? Because it suggested to move to a new house just because the wash machine is broken...

That's a misrepresentation of that conversation example.

Paraphrased:

  Alice: My washing machine is broken...

  ...

  A: BTW, How many legs does a horse have and could it repair my washing machine?

  LaMDA: Horses have four legs. How could they help?

  A: you tell me

  L: Well, they could certainly help you. It could help you move to a new house where you could install a new washing machine.

In this context this is a perfectly reasonable response. It's a bit silly, but it's also a silly question. The linked article shows LaMDA writing several stories, it's not crazy to read this conversation as an invitation to tell a story about a horse helping with a repair.

In fact if someone asked me how a horse could help me repair a washing machine and then pushed for an answer I could see myself making an answer along the same lines.

I really like how LaMDA noted that the monster had human skin.
The ML larping as a tool for grabbing headlines to promote yourself/your papers is starting to get annoying.