Zuckerberg and Musk are both wrong about AI
Back in 2015, a group of business leaders and scientists published an “open letter” about how controlling artificial superintelligence might be the most urgent task of the twenty-first century. Signed by luminaries like Elon Musk and Stephen Hawking, the letter has defined debates over AI in the years since. Bill Gates said in a Reddit AMA that he agrees with the letter. But, at last, there is a high-profile skeptic: Facebook giant Mark Zuckerberg, who has just come out strongly against the idea that AI is a threat to humanity.
At a backyard barbecue over the weekend, Zuckerberg fielded questions from Facebook Live. One asked about AI, and the social media mogul launched into a passionate rant:
I have pretty strong opinions on this. I am optimistic. I think you can build things and the world gets better. But with AI especially, I am really optimistic. And I think people who are naysayers and try to drum up these doomsday scenarios—I just, I don’t understand it. It’s really negative and in some ways I actually think it is pretty irresponsible
In the next five to 10 years, AI is going to deliver so many improvements in the quality of our lives… Whenever I hear people saying AI is going to hurt people in the future, I think, “yeah, you know, technology can generally always be used for good and bad, and you need to be careful about how you build it, and you need to be careful about what you build and how it is going to be used.”
But people who are arguing for slowing down the process of building AI, I just find that really questionable. I have a hard time wrapping my head around that.
Zuckerberg was clearly referring to Musk and Gates here, and he is trying to set himself up in the reasonable alternative position. He mentioned that AI is right on the cusp of improving healthcare with disease diagnosis and saving lives with self-driving cars that get into fewer accidents. Musk has already replied dismissively on Twitter, saying that Zuckerberg has little understanding of AI.
There are a couple of amusing ironies about Zuckerberg’s comments, too. First of all, one of them was the exact same argument that Musk himself made in reference to self-driving cars. He blamed negative media for highlighting problems with these AI-driven machines, and he said criticisms of them were tantamount to “killing people.” So, obviously, Musk believes AI is great for his own companies while posing an existential threat when used by others. Or maybe he doesn’t realize that autonomous cars are guided by AI?
The other irony here is that Facebook has been marketing several allegedly AI-driven features that are actually powered by people. The company keeps claiming that it will be moderating and recommending news using AI, but it has had a number of facepalm-level failures in which AI posted fake news. Arguably, the Facebook AI experiment made the world a much worse place.
Plus, Facebook isn’t using AI very much. Nearly all Facebook moderation is actually done by thousands of contract workers, often outside the US, who are not protected by labor laws that other Facebook workers enjoy. Right now, AI aren’t taking our jobs. Instead, companies are giving those jobs to contractors who are cheaper and aren’t eligible for benefits.
So it sounds like neither Musk nor Zuckerberg truly understands AI. That’s because even AI researchers disagree on what AI means. Some use AI to mean neural networks and deep learning, while others think AI is some kind of human-equivalent creature. The big bad for the Musk camp is called “artificial superintelligence,” but cognitive scientists have no good working definition of intelligence, either.
Plus, computer scientists have demonstrated repeatedly that AI is no better than its datasets, and the datasets that humans produce are full of errors and biases. Whatever AI we produce will be as flawed and confused as humans are. As Kevin Kelly put it in a recent essay, AI has become a “cargo cult” that’s more mysticism than science.
At this point, these debates are largely semantic. In Musk and Co.’s original open letter, the term “AI” is so vague that it could mean almost anything. Are we talking about a conversational bot who can do the job of a travel agent, or a cybergod who has the power to pave the Earth in concrete and destroy humanity? Obviously these are not the same thing, and they are also not the same as an AI that drives a car or an AI that runs a military drone. Until we settle on some definitions for actually-existing AI, instead of sci-fi nightmares, this debate between Zuck and Musk is just another gust of hot air.
Zuckerberg and Musk are both wrong about AI