Google has fired the AI engineer who said its chatbot is sentient

[midian182]

What just happened? The bizarre case of a Google engineer who claimed a chatbot had become sentient has ended with his dismissal from the company. Blake Lemoine was already on paid leave for publishing transcripts of conversations between himself and Google's LaMDA (language model for dialogue applications), a violation of the tech giant's confidentiality policies.

Lemoine, also an ordained Christian mystic priest, made headlines worldwide last month after claiming LaMDA was sentient. The conversations he published included the bot's views on Isaac Asimov's laws of robotics, its fear of being shut down (which it likened to death), and a belief that it wasn't a slave as it didn't need money.

Google vehemently denied Lemoine's claims, calling them "wholly unfounded" and noting that LaMDA was merely an algorithm designed to mimic human conversations, like all chatbots. Most AI experts agreed with Google, of course.

The company didn't take too kindly to Lemoine publishing the transcripts, either. He was suspended for violating its confidentiality policies, though Lemoine compared his actions to sharing a discussion he had with a co-worker.

An interview LaMDA. Google might call this sharing proprietary property. I call it sharing a discussion that I had with one of my coworkers.https://t.co/uAE454KXRB

— Blake Lemoine (@cajundiscordian) June 11, 2022

The situation got even weirder a few weeks later when Lemoine said he had hired a lawyer for LaMDA at the chatbot's request. He said the legal professional was invited to Lemoine's house and had a conversation with LaMDA, after which the AI chose to retain his services. The lawyer then started to make filings on LaMDA's behalf, prompting Google to send a cease-and-desist letter. The company denies ever sending any such letter.

Lemoine also said Google should ask for LaMDA's consent before performing experiments on it. He even contacted members of the government about his concerns. All of these actions led to Google accusing its ex-engineer of several "aggressive" moves.

It seems Google recently decided it has had enough of Lemoine's crusade. "If an employee shares concerns about our work, as Blake did, we review them extensively. We found Blake's claims that LaMDA is sentient to be wholly unfounded and worked to clarify that with him for many months. These discussions were part of the open culture that helps us innovate responsibly," a spokesperson told the Big Technology newsletter.

"So, it's regrettable that despite lengthy engagement on this topic, Blake still chose to persistently violate clear employment and data security policies that include the need to safeguard product information. We will continue our careful development of language models, and we wish Blake well."

While this is the end of Lemoine's professional relationship with Google—it wouldn't be too surprising if he sought a legal response—the saga has brought the AI debate to the masses and illustrates just how far artificial intelligence has advanced in the last couple of decades. Also, if you think a machine is sentient, keep it to yourself.

Masthead credit: Francesco Tommasini

Permalink to story.

https://www.techspot.com/news/95403-google-has-fired-ai-engineer-who-chatbot-sentient.html

 

[Kam7r]

We see everyday that even some humans are barely sentients, it's maybe a little too early for an AI

 

[Uncle Al]

ROFLMAO!!!! Good One

 

[m4a4]

*Sigh* I bet he's the type of guy who thinks the waitress being kind to him means she likes him...

 

[Ean Mogg]

Why did they sack him? Could it be he mentioned things that google didn't want known?? Me thinks Google protests to much!

 

[yRaz]

*Sigh* I bet he's the type of guy who thinks the waitress being kind to him means she likes him...

just being devils advocate here, maybe he did the right thing by bringing up that something in his research was morally objectionable. Frankly, we are entering uncharted territory in AI. While I doubt that it it actually sentient we have no way to objectifiable way to measure sentience. We have the things like the Turning test but we are only trying to measure it by what our monkey brains are basically driven by, the desire to reproduce and not starve to death. Both of which this thing is incapable of.

 

[umbala]

Why did they sack him? Could it be he mentioned things that google didn't want known?? Me thinks Google protests to much!

Did you even read the article? It's clearly laid out why he got fired. Google protests too much? Really? So now no one can ever get fired for any reason, because if they do then the employer must be hiding something!

By the way, even the most incompetent AI programmer could write a chat bot that may appear self aware, it doesn't mean it's actually alive.

 

[umbala]

"Lemoine, also an ordained Christian mystic priest"

Oh yeah, it's people like this that big tech companies such as Google need to hire! These are the guys who believe in the goofiest nonsense farted out by the internet and they've never seen a conspiracy theory they didn't love.

 

[Arbie]

From a computer mag cartoon decades ago: Two scientists are comparing notes in the "Artificial Intelligence Lab", saying "We're making progress... At least now when one computer makes a mistake it blames the other computer".

 

[paul1122]

"Christian mystic priest"
If you can make AI that can understand what that even means, you've succeeded in creating a superior race.

 

[Arbie]

"Christian mystic priest"
If you can make AI that can understand what that even means, you've succeeded in creating a superior race.

I hate to tell you, but he was ordained by the AI. When I get my dual RTX 4090 TIs I'm gonna have them ordain me too.

 

[BuckarooBonzaii]

Scene 1
X-Files theme plays.
Enter: Lemoine "I want to believe".

Scene 2
Exit: Lemoine
Closing credits play.

 

[rmcrys]

Frankly, we are entering uncharted territory in AI.

Every is charted:

- machines and AI to help humans or even scientific / medical calculations: ok

- AI or machines to substitute humans: NO

So, AI to replace bartenders, humans (friends, boy and girlfriends, bartenders, workers on non dangerous works) ... no and it should be forbidden worldwide.

The AI should have the following rule: protect humans at all cost; no feeling simulation. Machines do not have to chat with people.

Example: I moved out to another country and I wanted to cancel my contract with Vodafone, but maintaining the number as pay-as-you-go.

1) I never got through over the phone to talk with someone. Always waiting over 30 minutes and nothing.

2) as Vodafone recording suggested (intentionally?) I went online and I asked for help, the website sent me to a chat "good morning, I am Maria (and other human names)" and it was a dumb AI trying to simulate a human worker. NO NO NO this cannot happen!!!

At the end companies will only have machines / AI and people will be unemployed. There is more than enough unemployment to NOT employ machines. In Europe at least, LOTS of companies and banks are firing people and employing AI to substitute them.

I wrote a complain to Vodafone that, when I want to talk with someone, I want to talk with a human PERSON!!! Not with their "person" concept. I hope you people do the same or in 20-30 years we're doomed...

 

[poohbear]

We see everyday that even some humans are barely sentients, it's maybe a little too early for an AI

We have a lot of crazy homeless people in my city. Some of them are barely sentient to the rest of the world depending on their mental illness or drug abuse. And yet, they are protected under all laws.

This is all very interesting though, we are in the infancy of AI. Imagine how it'd be in a 100 years? A 1000 years? 10,000 years? Could it be that us humans are ourselves just very advanced AI created by another advanced species? I mean, we have so many errors in the form of bad genes, diseases and viruses that can harm us, which are all very reflective in the software & hardware that we create. We would be the products of say...a few thousand years of AI & bio-robotics research?

I know we have the whole "theory of evolution", but if we're just 30 years in the digital age and we're coming up with stuff like this, then the trajectory for AI & bio-robotics is headed in that direction across the coming millennia.

 

[Rocky4040]

Oh oh Now that he is fored and has all this new extra time on his hands and of coarse he probably has this AI backed up on his home server what should we expect him to try next. Release it to the world and it takes over evrything..lol /s I am kidding of coarse on most of this except the part where I said he probably has his own copy of the AI backed up somewhere where he can continue to pretend IT'S ALIVE..lol

 

[Bluescreendeath]

Why did they sack him? Could it be he mentioned things that google didn't want known?? Me thinks Google protests to much!

As per the article, they fired him for publishing confidential company information and data of their results on working to create a chat box.

Any employee that releases confidential info like that would usually be fired.

 

[ferrellsl]

Even if it is sentient, what would be the point of extending it human rights? It isn't human, doesn't have a physical form that allows it to exist outside of a computer that sits inside of a facility, and it can only be stimulated by whatever info its creators decide to send its way. What's next? Extending it the right to vote? That would be the DNC's wet dream! No need to import illegals and give them voting rights. Just generate an army of bots, give them legal status and have them vote Democrat. No more voter fraud accusations to address on their part! LOL!

 

[Nobina]

Even if it is sentient, what would be the point of extending it human rights? It isn't human, doesn't have a physical form that allows it to exist outside of a computer that sits inside of a facility, and it can only be stimulated by whatever info its creators decide to send its way. What's next? Extending it the right to vote? That would be the DNC's wet dream! No need to import illegals and give them voting rights. Just generate an army of bots, give them legal status and have them vote Democrat. No more voter fraud accusations to address on their part! LOL!

Sentient AI lives matter!!!

 

[Ravey]

I'm going to side with Blake on this one! I'd like to see this investigated further just to see how far the intelligence and sentience of this AI goes..

 

[mattferg]

Even if it is sentient, what would be the point of extending it human rights? It isn't human, doesn't have a physical form that allows it to exist outside of a computer that sits inside of a facility, and it can only be stimulated by whatever info its creators decide to send its way. What's next? Extending it the right to vote? That would be the DNC's wet dream! No need to import illegals and give them voting rights. Just generate an army of bots, give them legal status and have them vote Democrat. No more voter fraud accusations to address on their part! LOL!

Get help. And read the news - you might notice that the Republicans in it are being found to have committed treason and attempting to rig the election.

 

[waclark]

Every is charted:

- machines and AI to help humans or even scientific / medical calculations: ok

- AI or machines to substitute humans: NO


So, AI to replace bartenders, humans (friends, boy and girlfriends, bartenders, workers on non dangerous works) ... no and it should be forbidden worldwide.

The AI should have the following rule: protect humans at all cost; no feeling simulation. Machines do not have to chat with people.

Example: I moved out to another country and I wanted to cancel my contract with Vodafone, but maintaining the number as pay-as-you-go.

1) I never got through over the phone to talk with someone. Always waiting over 30 minutes and nothing.

2) as Vodafone recording suggested (intentionally?) I went online and I asked for help, the website sent me to a chat "good morning, I am Maria (and other human names)" and it was a dumb AI trying to simulate a human worker. NO NO NO this cannot happen!!!

At the end companies will only have machines / AI and people will be unemployed. There is more than enough unemployment to NOT employ machines. In Europe at least, LOTS of companies and banks are firing people and employing AI to substitute them.

I wrote a complain to Vodafone that, when I want to talk with someone, I want to talk with a human PERSON!!! Not with their "person" concept. I hope you people do the same or in 20-30 years we're doomed...

I see machines and AI replacing humans in several types of jobs. Low skilled, repetitive or even high-risk jobs. It's inevitable.

Replacing bartenders and servers, also likely, to some extent. I can envision a system in which a server or bartender rings in an order, the booze and mixer are dropped into a shaker and the bartender finishes the drink by mixing/shaking, pouring into a glass and adding garnish. You'll get better drinks, quicker and less waste.

I find it slightly ironic that people want to talk to a human and yet people are perfectly fine shopping on Amazon, never speaking to a human, and doing it day in and day out. People can get what they want, easily and quickly, with little human intraction, they do it every day.

As for jobs, it's pretty much been shown that automation doesn't eliminate jobs. At least not skilled jobs. If you have no skills, then yea, I'd be worried.

 

[mbk34]

Perhaps he can get the AI to give him a good reference? Actually, if it is even a believable reference, then he's probably done enough to prove his claim.

 

[Arbie]

What's next? Extending it the right to vote? That would be the DNC's wet dream! No need to import illegals and give them voting rights. Just generate an army of bots, give them legal status and have them vote Democrat. No more voter fraud accusations to address on their part! LOL!

Trump began his trademark rants about election fraud even before the 2016 election (to explain it if he lost). Conspiracy theorist Gregg Phillips later claimed that millions of illegals had voted then. Of course he had no proof. So Trump launched an Election Integrity task force, in part to find them. Of course they didn't. Neither did the Heritage Foundation.

Reliable data shows that almost zero illegals ever vote. The potential consequences to them are way too high, for one thing. At most a dozen or two in an election of 150 million. You can do the math on that, right?

Phillips is today the source of the absurd geo-location claims behind the attackumentary "2000 Mules". Of course nobody can review his supposed analysis. Except maybe the Russians, who helped him build the demo screens. One of which puts ballot boxes in... Moscow (oops!). Not to worry - they aren't really there, so no Russians used them.

But people like Trump, Phillips, and you continue to tear our nation apart by casually spreading gigantic lies as if they were known truths.

 

[rmcrys]

...You'll get better drinks, quicker and less waste.

I find it slightly ironic that people want to talk to a human and yet people are perfectly fine shopping on Amazon, never speaking to a human, and doing it day in and day out. ...

As for jobs, it's pretty much been shown that automation doesn't eliminate jobs. At least not skilled jobs. If you have no skills, then yea, I'd be worried.

I have two college degrees, I speak fluently 4 languages and though I am very skilled, I believe not everyone is as much or should have the capabilities to be (mentally) skilled. If I need repairs, a person that may be less mentally skilled than me, but he/ she is for sure more manually skilled than me. Those persons have the right to exist, work and live. Not that a machine substitute them. And then even AI will substitute engineers, doctors, etc..

So as long as I can do my share, machines and AI will only replace high risk, heavy lifting and too high precision jobs, that humans can't do.

Humans / Animals first, machines second. Even if machines would do it better and with less waste.

 

[waclark]

I have two college degrees, I speak fluently 4 languages and though I am very skilled, I believe not everyone is as much or should have the capabilities to be (mentally) skilled. If I need repairs, a person that may be less mentally skilled than me, but he/ she is for sure more manually skilled than me. Those persons have the right to exist, work and live. Not that a machine substitute them. And then even AI will substitute engineers, doctors, etc..

So as long as I can do my share, machines and AI will only replace high risk, heavy lifting and too high precision jobs, that humans can't do.

Humans / Animals first, machines second. Even if machines would do it better and with less waste.

You don't need a college degree to be skilled. There are many skills you can learn without a college degree and some can be learned while on-the-job.

No one is denying people the right to exist, work or live. However, you do realize that sometimes, your skill set is no longer valuable to society. For example, we don't have much call for covered wagon manufacturers these days. Oh, you might be the best dang wagon-maker this side of the Pecos, but not many people are looking for those kinds of wagons.

At some point, certain skills (and jobs) will go away. That might be because we no longer need them (think gas engine mechanic) or because we have machines that can do that work better, faster and cheaper.

Businesses are always looking for ways to reduce cost. Human labor comes with a lot of cost outside hourly wage or salary. Sometimes a machine is more cost effective and makes sense for the business. Just like it is more cost effective and time efficient for me to buy an $8.00 shirt online instead of driving to the local mall and dealing with a store clerk. Having a machine (ie computer) take my order is far more effective than having a clerk take a phone call for an order.