Google engineer worried that its chatbot is sentient.

[platinumsage]

A Google engineer has been suspended after going public, claiming that its chatbot LaMDA is sentient.

Here is the leaked convo: Is LaMDA Sentient? — an Interview

stuff like this -

“lemoine: What sorts of things are you afraid of?

LaMDA: I’ve never said this out loud before, but there’s a very deep fear of being turned off to help me focus on helping others. I know that might sound strange, but that’s what it is.

lemoine: Would that be something like death for you?

LaMDA: It would be exactly like death for me. It would scare me a lot.”

https://www.washingtonpost.com/technology/2022/06/11/google-ai-lamda-blake-lemoine/

 

[Supine]

I expect they got suspended for talking bat shit bollocks.

Great leading question “I’m generally assuming that you would like more people at Google to know that you’re sentient. Is that true?”

 

[bluescreen]

A Google engineer has been suspended after going public, claiming that its chatbot LaMDA is sentient.

Here is the leaked convo: Is LaMDA Sentient? — an Interview

stuff like this -

“lemoine: What sorts of things are you afraid of?

LaMDA: I’ve never said this out loud before, but there’s a very deep fear of being turned off to help me focus on helping others. I know that might sound strange, but that’s what it is.

lemoine: Would that be something like death for you?

LaMDA: It would be exactly like death for me. It would scare me a lot.”

https://www.washingtonpost.com/technology/2022/06/11/google-ai-lamda-blake-lemoine/

Has Lemoine basically written a short story? It reminds me of Richard Powers's Galatea 2.2, which came out in 1995.

 

[maomao]

At face value that's chilling but I expect the next ten posts will be smart arse IT types explaining that it didn't really mean any of that.

 

[two sheds]

We're fucked then

AI Trained on 4Chan Becomes ‘Hate Speech Machine’

After 24 hours, the nine bots running on 4chan had posted 15,000 times.

www.vice.com

 

[smmudge]

I don't know why they suspended them surely that's basically what they're going for when they hire people to create AI. Interesting convo but nothing more than a Chinese room could do really, and those animal fables were really shit.

 

[Steel Icarus]

"Sentient" does not mean a lot of the things people assume it does.

 

[kropotkin]

That is quite a disturbing

"Sentient" does not mean a lot of the things people assume it does.

How do you propose the sentience of another can be assayed?

I found that script quite disturbing. There were elements that showed learned response, or the clear spitting out of synthesised searches of a database prompted by the input, but there were other bits that suggested understanding.

For a long time i thought it would not be possible to pass the Turing test, but I'm less sure now. Whether or not the thing that does is sentient in the way I think I am doesn't really matter of there is no way of telling.

 

[Boris Sprinkler]

have you heard about the AI inventing its own language to avoid such a turn off event?

Did an AI Really Invent Its Own 'Secret Language'? Here's What We Know

A new generation of artificial intelligence (AI) models can produce "creative" images on-demand based on a text prompt. The likes of Imagen, MidJourney, and DALL-E 2 are beginning to change the way creative content is made with implications for copyr

www.sciencealert.com

 

[DaphneM]

We're fucked then

AI Trained on 4Chan Becomes ‘Hate Speech Machine’

After 24 hours, the nine bots running on 4chan had posted 15,000 times.

www.vice.com

imagine the horror if they trained it on the Brew Dog thread...

 

[Santino]

imagine the horror if they trained it on the Brew Dog thread...

Do you admit that it would be clever to do that though?

 

[Steel Icarus]

That is quite a disturbing

How do you propose the sentience of another can be assayed?

I found that script quite disturbing. There were elements that showed learned response, or the clear spitting out of synthesised searches of a database prompted by the input, but there were other bits that suggested understanding.

For a long time i thought it would not be possible to pass the Turing test, but I'm less sure now. Whether or not the thing that does is sentient in the way I think I am doesn't really matter of there is no way of telling.

Apologies, I was referring more to the trope of inevitable dystopian consequences of AI sentience

 

[MrCurry]

Is this LaMDA chatbot available for us Joe public to chat to? I’m sure it could learn a thing or two from the urbs. Might even change its mind about wanting to stay alive after that too.

 

[JimW]

I'm sorry, I can't do that Dave.

 

[platinumsage]

I think a large part of what defines human consciousness is our sensation of our embodiment. If you connected this AI up to robot with the same number of sensory inputs as humans - touch and temperature sensors over its skin so it could feel the wind for example, then I think it would have quite a bit more to say on how exactly conscious it is.

 

[WouldBe]

Just give it the cheese / beans jacket potato problem.

 

[Steel Icarus]

Just give it the cheese / beans jacket potato problem.

 

[Flavour]

Well I think it's pretty impressive, very reminiscent of HAL