Google's rush to win in AI led to ethical lapses, employees say

[Jan]

Google employees call company's ChatGPT rival a "pathological liar" and "worse than useless"

Shortly before Google introduced Bard, its AI chatbot, to the public in March, it asked employees to test the tool.

One worker’s conclusion: Bard was “a pathological liar,” according to screenshots of the internal discussion. Another called it “cringe-worthy.”

 

[DapperDanMan]

"Don't be evil!"
"Whoops ... too late!"

 

[1am7h30n3]

It's certainly not perfect, but it is fun to play with and it's always nice to have another option for LLM.

I did find that it's too eager to agree. I asked it about PMB health benefits, which is data that can be looked up and verified online. When I would say "I don't think it's covered", it would agree that it's not covered. Then I would say something like "oh no, I was on the wrong page, here it is, seems like it's covered", and it would agree that it's covered.

It also doesn't seem to be able to do simple things like ROT13 text encoding/decoding, or converting to/from ascii values, even though it's adamant that it can.

 

[FiestaST]

I’ll buy that.

 

[Vorastra]

It wasn't completely woke enough for Google's employees.

 

[_TrXtR_]

Don't know what they so worried about Google responses. ChatGPT is full of bad advice, ignoring given facts when asked to look at the facts... things like that. I'm pretty sure it can get you killed as well.

 

[Pak Fa Fui]

Cant install man made ethics into a machine