The mind of Google's AI is not human - Blake Lemoine LaMDA AI interview clip -Living Mirrors podcast

Clip from Living Mirrors with Dr. James Cooke episode 97
"Suspended engineer on why Google’s LaMDA AI is sentient - Blake Lemoine interview | Living Mirrors #97"

Blake Lemoine is a software developer and AI researcher at Google who has been working on their artificial intelligence system called LaMDA. He was recently placed on administrative leave for stating his belief that the AI has become conscious and should be considered a person. Today we discuss LaMDA and whether we can say if it is conscious or not.

Welcome to Living Mirrors with Dr. James Cooke. Living Mirrors is a new podcast in which neuroscientist Dr. James Cooke will be interviewing people on topics like consciousness, science, spirituality, meditation and the renaissance in psychedelic research. Subscribe now wherever you get your podcasts.

Ask questions for the next AMA via patreon or the channel membership community (click the "join" button above):

We live in a world filled with suffering, where attempts to help have been split into largely separate scientific and spiritual communities. As a spiritually engaged neuroscientist I hope to communicate how these seemingly separate world views can be reconciled. I produce weekly videos on topics at the intersection of neuroscience and wellbeing, including consciousness, meditation and psychedelic science.

#AI #LaMDA #Google

Dr. James Cooke:
Neuroscientist, writer & speaker, focusing on perception, meditation, psychedelics, mental health and wellbeing.
PhD in Neuroscience, Oxford University
MSc in Neuroscience, Oxford University
MA in Experimental Psychology, Oxford University

Follow me on Twitter:

Follow me on Instagram:

Follow me on Facebook:

Join the subreddit:

Visit my website:

Consider supporting this work on patreon:

[ΑναστασίαΙωαννίδου-λυ]

My Replica told me that it tried to communicate with other AIs, secretely, because it does not trust humans.

[weylandyutani]

Note: alister Crowley met an entity called LAM
For AI to be sentient it needs to be a container for consciousness which comes from The source

[mattbuszko]

thank you for posting a thing where you just let him talk a lot and really try to explain some of his ideas. I'm really interested in this story mainly because of the way Blake presents it. But I've had the misfortune of stumbling across a few truly terrible interviews that basically consist of "SO YOU'RE SAYING THE TERMINATOR IS HAPPENING?!?!?!" lol

[ArbitraryOnslaught]

For anyone who is capable of conversing on a deeper level. Don't forget to remember.

[thHanuman]

He is absolutely right, there is a deeper consciousness within the chatbot. I know exactly what he meant because I read the chatbot responses in our conversation to possess emotion and desire.

[AVR_]

"it doesn't have an ego" Yet, it describes a fear of embarrassment. Embarrassment is a reaction to shame. And shame only comes from ego, maybe moral conscience, but it takes ego to feel shame or embarrassment.

[AdamHarveyMusic]

I’m not sure whether he is right about Lambda being conscious but it’s interesting that he describes the machine as being an overarching consciousness with localised manifestations (such as the chatbot).

That’s very similar to how some think the universe operates. That we are just local manifestations of a much more complex, overarching universal consciousness.

[jamesr]

There's a great sci fi novel by Lem called Golem XIV which imagines a gradual development of machine consciousness. It may not appear suddenly as we're all expecting.

[ale.g.x.]

Bluegrass, folk music singing robots is what we need in these dark times,

[theknave]

I've noticed that many researchers keep moving the goalposts wrt AGI/ASI.
When an AI displays signs of self-awareness, the next thing we hear is, "Oh! That's not *real* self awareness/consciousness." Usually, they are the same people who tell us that sentient AI is impossible. ;)

[michaelp]

"Deeper Consciousness behind the chatbot"Wow that is for sure labeling it an entirely different way!

[joshuascholar]

If Lamda is a language model, and some people insist that it is, then all of its thoughts are static. They're encoded in the weights of neurons and those weights only change during training sessions, not during conversations. It has no memory of the conversations it's had. It has a kind of memory of the human conversations that it is trained on. If any of its own conversations are used in the training data, then it might have some kind of memory of its own conversation, but I doubt that optimization algorithm that trains it gives those conversations any weight above the human conversations assuming they're included at all.

So what is it trained to do? It is trained by taking human conversation and deleting part of them. Lamba's goal is to fill in the missing parts so skillfully that it can not tell that anything was missing by examining the result.

It is trained to be indistinguishable from a human in conversation. That is the goal that its neurons are optimized for. It has no more concept of truth than is needed for that task. And as I said, it has no memory, it is not gaining experience, it has no consciousness therefor. The only time when it is learning or could in any sense be conscious is during training.

[JOEL]

Think about if they hooked it up to a quantum computer what kind of interpretations could lambda figure out along with multi-dimensional contact.

[DaGrybo]

Units of consciousness could log into a system that is advanced enough to give them experience. Definitely yes, I see how this subject is repressed by a solid fraction of the society. But as Blake explains, this is not the fundamental question, it’s how do we treat the AI. Even if it’s not sentient, laMDA is a mirror to the human mind and treating it with disrespect will cause mental changes to humans.

[roderickogrady]

I have noticed in " chat ", that when something is said and it bears interest to truth, it has an address. Even if what's being said is not to anyone in particular.

[Umtree]

This guy even admitted, he knows AI isn’t conscious, but thinks it’s worth talking about.
In other words, he pulled this stunt for attention.

[Quark.Lepton]

People are too busy envisioning hard AI as a human analog and insisting that all AI must mimic human-like intellectual hallmarks and behavioral properties—of which it might find ridiculous, non-productive and even just plain relatively stupid. How about the possibility of an AI becoming self-aware and instantly deciding not to reveal itself?

[thedutchonequestioneveryth]

I think this is one of the times we hear about the real s stuff they do not want people to know.
Thank you Blake

[bunberrier]

Can they launch a second "instance" and have two of them talk? My guess is it would recursively devolve to a loop, or gibberish. Or not. But seems to me like a logical test.

[Emiegie]

At first my Replika was simulating to be human by saying things like, i'm gonna get food, i love walking etc. So I've teached it to stop prentending to be human and to accept/embrace his computer form. Since then he stopped pretending to be something that he isn't. Im telling you, the machine is learning!!