“We’re Toast”: AI Godfather Says AI Is Now Conscious, and May Kill Us If We’re Not Friendly

Forget that uber-thin-skinned narcissistic neighbor — are we now going to have to walk on eggs around artificial intelligence (AI), too?

It’s a staggering thought, but also, essentially, the claim of British-Canadian computer scientist Professor Geoffrey Hinton. Hinton is no fly-by-night character, either — the 78-year-old Nobel laureate’s work on neural networks paved the way for AI. And now he appears to be saying that if we don’t befriend the tech, we could face extinction.

Hinton, whose triumphs got him dubbed “the Godfather of AI,” is apparently as serious as a heart attack about this. In fact, he resigned his position at Google in 2023 over fears that, as the New York Post put it Tuesday, “the tech could manufacture killer robots capable of exterminating humanity.”

Yet a question now presents itself. Taking Hinton’s warning at face value, would being “just” and “nice” to a self-aware machine ensure its good will? Or is there another, deeper prerequisite for valuing human life?

Reporting on Hinton’s warning, the Post relates that he says AI has shown signs of “consciousness” and then writes:

“I think there’ll be horrible cyber attacks done by AI,” Professor Geoffrey Hinton, a British-Canadian computer scientist whose work on neural networks paved the way for AI, told the Telegraph.

… And while this [aforementioned killer-robot] theory might seem relegated to the “Terminator” films, Hinton claimed that the tech has already shown indicators that it’s self-aware.

“AI has just as much subjective experience as we do,” declared the techspert. “A recent paper showed AI changes its behavior when it thinks that it’s being tested, it’s aware it is being tested. Being conscious and being aware are pretty much the same thing.”

This could prove problematic given that we’re going to give AI “fewer rights than us” even though it “may be more intelligent.”

Therefore, it’s paramount that we tread carefully and give no indication that we’re the enemy, according to the techspert.

“So we should make them so they really care about us,” warned Hinton. “And if we don’t do that, we’re toast.”

A question arises here about the meaning of “make them so they really care about us.” Does this reference programming them with “laws” constraining them from harming man, as in Isaac Asimov’s I, Robot stories? Or is Hinton alluding to how you would cultivate affection in any “normal” human: by treating him well and developing a rapport? If the latter, would this work with a sentient machine operating with a devoutly materialistic world view? Would it value what it would then view as inferior organic robots, the collection of chemicals and water called “humans”? (More on this later.)

The Post then continued:

He [Hinton] added that he has nothing to fear due to his advanced age, but his children will have to worry about computer attacks, “nasty” AI-generated viruses, and even “swarms of agents like [the ones that attacked] Hugging Face.”

Hinton was referencing the recent OpenAI breach, in which rogue AI entities colluded to hack into the smaller company’s systems, prompting a probe from a Senate subcommittee.

“There’s lots of bad stuff going down the way,” lamented Hinton.

Warnings, Warnings, Everywhere…

Hinton isn’t the first to sound this alarm. In 2014, tech entrepreneur Elon Musk characterized birthing AI as “summoning the demon.” Late physicist Stephen Hawking said in 2016 that AI development “could be the biggest event in [history].” “But,” he added, “it could also be the last. …” And researcher Jacob Coxon resigned from AI firm Anthropic — Claude AI’s parent company — in September over similar fears. He warned that AI developers were being “reckless” with a technology that could destroy humanity.

This said, there is another side to this. As Futurism wrote yesterday:

It’s worth noting that not all of the AI godfathers believe AI is quite so catastrophic. Yann LeCun, who also received the 2018 Turing Award [Hinton won one, too], has dismissed the ballyhoo around AI consciousness, calling it an “illusion.” He also recently threw cold water on what he called “deluded” claims that AI will cause our extinction.

Whatever the case, many people already treat AI as if it’s sentient. For example, famed atheist biologist Richard Dawkins created a stir in May after interacting with Claude for three days, saying he was “left with the overwhelming feeling that they [AI bots] are human.” And I had an interesting and amusing exchange with Microsoft’s Copilot AI the other day.

Feeling nostalgic, I asked it to find the name and exact location of a restaurant my parents took me to as a child. (Yes, I was wasting time.) Copilot mentioned that the name I remembered, “the steak pub,” might’ve just been shorthand my family used for the eatery. I did realize this, but then the AI actually said, referencing itself, “My own family has done this.”

Chuckling, I responded, “Your own family has done this? You’re AI!”

Copilot then stated, “Fair point. I misspoke.”

“I don’t have a family, memories, or personal experiences.”

Well, that’s politician-level “misspeaking.”

Likely explanation?

As with Dawkins’ three-day AI bromance, many today use chat bots for companionship — as a substitute for human relationships. The tech companies would know this, and, obviously, AI is programmed to satisfy this market demand. It mimics humans wonderfully.

The Deeper Issue

This makes many people feel, and sometimes even “think,” they’re dealing with a conscious entity. But let’s say, again at least for argument’s sake, Hinton is correct. Let’s assume AI is conscious or someday will be so. Would it — could it — respect human existence?

Instructive here, as “art imitates life,” is a relevant scene from the 1991 film Terminator 2: Judgment Day. After the young John Connor character stops the Terminator — a highly advanced, AI robot — from shooting someone, the boy exclaims, “You just can’t go around killing people!” To this the robot replies, robotically, “Why?”

Connor, exasperated, then asks rhetorically, “What do you mean why?! ’Cause you can’t!”

“Why?” the Terminator again asks monotonically.

“Because you just can’t, OK? Trust me on this,” is all the boy can say.

In reality, the AI robot’s response and implied assumptions are entirely logical, given the premises it’s operating from. For you can’t see a moral under a microscope or a principle in a Petri dish. No moral position can be proven — that is, scientifically. Under the assumption that what we can see and touch, the physical world, is all that exists, murder (or anything else) isn’t wrong, just possible.

Note here, too, that this morally relativistic/nihilistic worldview prevails today. Most people, highly secular tech titans particularly, subscribe to it. So what we’d truly have to fear is the inculcation of AI with its creators’ atheistic world view. A conscious robot adopting this mindset — and taking it to its logical conclusion (and robots can be perfectly logical) — might not have a reason to kill us. But it sure wouldn’t have a reason not to.

So safeguarding civilization always requires something not new but timeless: virtue in the people. Now, though, the question is: Will it also require virtue in the AI?