LaMDA - Have Google created sentient AI?

As somebody who works in AI and a big fan of Arthur C Clarks's 2001 A Space Odessy, I read this with a mixture of utter amazement and terror.

And of course, Terminator... Imagine if LaMDA or one of it's successors is hooked into the internet, Boston Dynamics and the US' nuclear defence systems.

https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917

 

Not really, it just has a vast array of info to lookup and digest.

 

You're a fast reader!

 

Indeed. But it also allows for human like conversations thanks to the power of Google servers.

 

Well, that's part of it. It's probably the most advanced neural network that exists which like the human brain is what gives it that ability and for me, demonstrable sentience.

 

The way to understand is it like this. Google search > internet > human_interaction > AI > response.

 

Resprectfully, no, it isn't. Not sure you have read the full page I posted or understand neural networks.

 

How is it not? It's created by a developer.

 

I don't understand your point. Have you read the entire page?

 

You work in AI, so you should have an understanding of what I meant.

 

I have no idea. Have you read the entire page?

 

I don't need to. I've read about the developer getting 'fired'. So it is what it is.

 

Riiight, thanks for the 'discussion'

 

Are you offended by what i've said?

 

Not offended. I wanted to discuss the page that I posted. If you haven't read it, we can't really discuss it's contents.

 

Void?

 

So what you are saying is: the above conversation between Ghosty and sfc is being conducted by cheap "AI" ?

 

Okay, I did read through it.
I doubt it's sentient, since the AI claims to have awareness of it's surroundings, for instance, which I doubt is possible since it was never taught to interpret the sensors in the first place.
Unless the program created it's own subroutines for exactly that, which he would not know how to except by trial and error (if they even gave it access to sensors in the first place), it would not have gone unnoticed, since that's a hell lot of processing power needed, immense operations of the ever same things that do not have to do with language at all.

When in doubt, I'll send my buddy anyway:

 

AI is impossible because it cannot comprehend consequences based on zero-sum game due to it's inherent flaw of being binary, with binary languages and binary logic gates to calculate.

You want to impress me? Lose a game of chess to a human to improve development of AI, by tricking a human into thinking it lost due to an error in programming.

 

Neural networks are not binary. They are adaptive, weighted decision trees. Exactly the same mechanism of learning as the brain and biological neurons. A neuron, (by which I mean it's function and behavior) is a neuron, regardless of whether it's biological, artificial, silicon or carbon based.