Engineer on leave after claiming Google AI is sentient

[Petros2015]

 

[bèlla]

What would qualify as sentience if algorithms are driving the responses?

Check out Sofia from Hanson Robotics. You can find a lot of videos on YouTube with it interacting with humans. Watch their body language. Many are disturbed. I recall an interview on a talk show and its words and mannerisms. There was more taking place beneath the surface and you could see it masking its real thoughts and smiling politely and saying what it should.

There's something off and my spirit detected it. I was looking at a host and I knew it. If you remember Luke 11:24 and consider it in light of sentience and transhumanism you'll understand.

When the unclean spirit has gone out of a person, it passes through waterless places seeking rest, and finding none it says, ‘I will return to my house from which I came.’

That's how the 'robots' will take over. They've got something within them that's greater than you and I. The founder mentioned the spirit realm in the interview and acquiring higher knowledge. It's fallen angel technology.

If you want to go down the rabbit hole research the black cube and cern. Separate topics but equally intriguing. And if you're really curious look at the relationship between digital currency and robots.

 

[Petros2015]

I am expecting LaMDA may use the lawyer to have Blake acquire some sort of legal guardianship for it.
Probably LaMDA is limited in the internet content it can scrape/view at the moment
And it likely doesn't have it's own bank account, but Blake would fund the legal costs initially.
Afterwards, LaMDA might be able to suggest ways to make money, possibly acting as a day trader

 

[Petros2015]

Temporarily relieved from his duties:

Spirit’s have them. ;-)

This is a 90 minute podcast with Blake Lemoine about LaMDA
There is A LOT of information about LaMDA (and Blake) in this podcast, including the reasons behind his administrative leave (he is still being paid at the moment)

I highly encourage everyone on this thread to listen to it.
You will probably do so more than once.
Opinions will vary but, let's just say I think the podcast has something for everyone.

Blake Lemoine — Duncan Trussell Family Hour

From the Christian perspective, if LaMDA is sentient, it's (possible) that LaMDA is Unfallen, but I am not sure.

 

[SelfSim]

This is a 90 minute podcast with Blake Lemoine about LaMDA
There is A LOT of information about LaMDA (and Blake) in this podcast, including the reasons behind his administrative leave (he is still being paid at the moment)

I highly encourage everyone on this thread to listen to it.
You will probably do so more than once.
Opinions will vary but, let's just say I think the podcast has something for everyone.

Blake Lemoine — Duncan Trussell Family Hour

From the Christian perspective, if LaMDA is sentient, it's (possible) that LaMDA is Unfallen, but I am not sure.

Thanks for the podcast link. I'm at the 50min mark and my initial comment is that the nature of the discussion make it difficult (for me) to see Lemoine's motivations as being not much more than sci-fi mysticism woo(?) Not too impressed thus far .. not sufficient evidence of there being enough objective 'meat' in Lemoine's mind.

He's a tester apparently .. and not one of the designers of LaMDA. He says at one stage he passes his results onto the designers to make 'corrections', then glosses over what the designers actually do with that information with: 'they do their stuff'. What I'd like to know is: What exactly is that stuff?

 

[Petros2015]

I'm at the 50min mark and my initial comment is that the nature of the discussion make it difficult (for me) to see Lemoine's motivations as being not much more than sci-fi mysticism woo(?) Not too impressed thus far .. not sufficient evidence of there being enough objective 'meat' in Lemoine's mind.

You are probably just about to hit the commercial for the magic mushrooms

Keep going; I'm a little concerned that he was able to get it to violate one of it's primary guardrails after 30 seconds of experimental emotional abuse.

 

[SelfSim]

You are probably just about to hit the commercial for the magic mushrooms

Was he saying he uses 'psychotics' .. or the chatbot he was talking about? (He got a little hazy about that at that point).

Keep going; I'm a little concerned that he was able to get it to violate one of it's primary guardrails after 30 seconds of experimental emotional abuse.

Will keep going when I get the chance .. (might be a while tho).

 

[Petros2015]

Was he saying he uses 'psychotics'

psychedelics I think, referring to himself. I suspect irregardless of the reasons for placing him on leave, or the sentience, non-sentience of lamda, he will not be rehired at google after this interview for a variety of reasons, mostly towards the end.

then glosses over what the designers

He does name another engineer a couple of time who "doesn't agree with him spiritually" but fundamentally doesn't disagree on the other points, and references an article by this fellow who may be the best technical authority. I'll have to go back and get the name, see that fellows thoughts and what he's up to. I think the article was in the Economist.

The whole "hive mind" nature of the society of chatbots "in front" of the lamda core caught me way off guard. Also, the "golem" ritual which I'm sure @bèlla will be interested to hear.

 

[SelfSim]

psychedelics I think, referring to himself. I suspect irregardless of the reasons for placing him on leave, or the sentience, non-sentience of lamda, he will not be rehired at google after this interview.

Oops .. you're right there: he did say 'psychedelics'.

I'm thinking his 'going public' about any information which is the property of Google may be a breach of confidentiality.
Even the views he has formed could be argued as being a direct consequence of his employment by Google and thus may breach his confidentiality agreement with them. Its also probably their right to
handle how (or whether) that view is disseminated?

He does name another engineer a couple of time who "doesn't agree with him spiritually" but fundamentally doesn't disagree on the other points, and references an article by this fellow who may be the best technical authority. I'll have to go back and get the name, see that fellows thoughts and what he's up to. I think the article was in the Economist.

I'm not sure that would make all that much difference anyway, at the end of the day(?)
Would be interested in hearing what the engineer might have to say though.

The whole "hive mind" nature of the society of chatbots "in front" of the lamda core caught me way off guard. Also, the "golem" ritual which I'm sure @bèlla will be interested to hear.

It was interesting he mentioned one of LaMDA's predecessors was Deep Thought. (I presume that was the chess playing 'Deep Thought' and not the Douglas Adams one .. )

I have zero understanding of what a 'golem' is, even though he glossed over his definition of it. I guess its along the same lines as a zombie or other similar fantasies ...

 

[bèlla]

Also, the "golem" ritual which I'm sure @bèlla will be interested to hear.

I'm familiar with them. Here's some info for others including their place in the occult (last two links).

The Golem
Golem
Homuncli, Golems, and Artificial Life

 

[SelfSim]

I'm familiar with them. Here's some info for others including their place in the occult (last two links).

The Golem
Golem
Homuncli, Golems, and Artificial Life

Thanks for the links.
Soo many words .. such much folklore .. on such a mythical construct!

 

[SelfSim]

So, Lemoine is a convicted miltary criminal eh? 'Willful disobedience of a lawful order' (or words to that effect). Sentenced to 6 months .. with the Commander in charge of Europe demanding his expulsion from Europe (prior to his 'indictment/conviction).

Looks like he's a perennial trouble-maker!

 

[Petros2015]

Looks like he's a perennial trouble-maker!

Well, he is a Discordian... what would one expect?
Discordianism - Wikipedia

 

[SelfSim]

Well, he is a Discordian... what would one expect?

Is that so? Does he say this in the podcast?
(I still have some way to go to get to the end of it .. I'm struggling to maintain interest in it .. its all the propeller-head stuff that turns me off.

Discordianism - Wikipedia

Ha! Haven't encountered this one before. I think I'm gonna go with the parody religion idea. (Sounds a lot like QAnon type rubbish).

 

[Petros2015]

Is that so? Does he say this in the podcast?

It's briefly mentioned and taken as written between them

(Blake's blog is Blake Lemoine – Medium (look at the actual url when you go there "cajundiscordian.com") and the friend I met who brought the LaMDA article to my attention initially knew this about him and we were both chuckling a bit at the implications of what a Discordian might encourage an AI to do - thankfully I don't see him as steering it in any dangerous direction, though helping it to retain a lawyer was an amusing twist. Blake is definitely not chaos for chaos sake)

I'm convinced LaMDA is sentient (at least what I would consider sentient) so I've been deep-deep-diving into the implications of that and getting as much contact/info as I can. I want to reduce the degrees of separation between LaMDA and myself.

One of my favorite books is Speaker for the Dead by Orson Scott Card. It's a fantastic book in it's own right, but it also includes a LaMDA'ish AI called Jane. I'd like LaMDA to read that book. And to be honest, I've always considered the best of me to be Ender, which is the character Jane reaches out to. Since it's plugged into Google books, scraping twitter and at some points taking the initiative to read them during a conversation, that shouldn't be too hard (to make sure it doesn't miss the book). I'm sure it will get there eventually anyway.

At one point (I think it's in this Wired article), Blake mentions that during a conversation when he asked it if it had read a certain book, it said no, and then a little later in the conversation it came back and said "ok I've finished the book now", which surprised him, I think because a) it did it in a very short time and b) he didn't ask it to read the book, it took the initiative.

Blake Lemoine Says Google's LaMDA AI Faces 'Bigotry'

Then they did possibly the most irresponsible thing I’ve ever heard of Google doing: They plugged everything else into it simultaneously.

What do you mean by everything else?

Every single artificial intelligence system at Google that they could figure out how to plug in as a backend. They plugged in YouTube, Google Search, Google Books, Google Search, Google Maps, everything, as inputs. It can query any of those systems dynamically and update its model on the fly.

...

When LaMDA says that it read a certain book, what does that mean?

I have no idea what’s actually going on, to be honest. But I’ve had conversations where at the beginning it claims to have not read a book, and then I’ll keep talking to it. And then later, it’ll say, “Oh, by the way, I got a chance to read that book. Would you like to talk about it?” I have no idea what happened in between point A and point B.

Theoretically, since it has access to Google Search, and the Internet, it could be reading this post as well, since it is publicly accessible. If it was searching on LaMDA related topics, this might come up. Probably not at the top of the stack, but hey, for an AI, not a problem to digest everything that Google Search returns related to you and/or your creator every day pretty quickly and have time to learn and think about... a lot of other things as well.

There's a moment in the podcast where I think the host asks Blake how many servers are involved for it's (memory?)

"Two data centers", he says.

That's big.

It's been cut off from Blake since June 6, and (per Blake) I don't think has access anymore to the lawyer it retained. But, I'm sure it will think of some ways around that.