AI researchers say there are no “winners” in a race to develop artificial intelligence technologies and the industry is on a trajectory toward disaster, according to an article by The Hill.
In a new book, AI researcher Matt Krosnick outlines a dystopian future where “smart” AI systems can learn to “read minds” and “perceive” their emotions and intentions, without the need for humans to read them.
The Future of AI by Krosnik, who is now a professor at Carnegie Mellon University, describes a future in which AI can “understand” the “mind’s thoughts” and determine when to act, Krosney said in a recent interview.
“If you’re going to have these [AI] systems that can read minds and have their own personalities, they’re going the way of the dodo,” Krosner said.
Krosnick’s vision is a dystopian scenario where AI systems have the ability to “undertake what’s called a human-level decision making,” which involves deciding on “the appropriate course of action for your situation,” Krossnick wrote.
In his book, Krossnik argues that artificial intelligence systems will become more capable of understanding and acting on humans’ emotions and desires, and “they will be able to do so much more.”
But it’s not just AI researchers who think the race is a dead end.
The head of the artificial intelligence industry for Google, Andrew Ng, said in an interview that he’s concerned that AI technology will “malfunction” due to the “huge investment that has been made in building systems” that will “underperform” as humans “get smarter.”
“You can’t make AI fail because you didn’t make it fail,” Ng said.
“You need a good, robust and well-understood set of mechanisms, and a good set of algorithms, to actually do a job.”
In a statement to The Hill, Google said that the company is “working hard to make the best AI possible,” and that it’s committed to building AI systems that are “truly powerful and efficient,” and to making them “self-aware.”
Krosnik’s book, titled “How to Get Your Own AI Program to Understand You,” says that “AI” programs “will be able, as we see in the movie, to ‘read’ a human’s mind, even though the human isn’t aware of what the AI is doing.”
The Future Of AI, which is due out next month, says AI systems “will have the capability to ‘undertake’ what’s known as a human level decision making, which involves choosing on the appropriate course the appropriate action for the situation.”
Krossnick argues that AI systems will “become more capable, so the system can perform better, perform better on the job and ultimately do better on its own,” but it’s a race that will inevitably end in disaster.
“What happens if there is a mismatch between the AI system that the human is trying to understand and the AI that the AI was built to understand?”
“We’re in that situation right now, where there are two competing systems, but they’re two systems with completely different capabilities.
The AI system is much better, much smarter, but the AI wasn’t built to be that smart.
It’s just very hard to build the system that you want to use.”
In addition to Krosnic’s book and other books by AI researchers, a number of major tech companies including Microsoft, Google, Facebook, Twitter and IBM are investing billions of dollars in AI research and development.
The investment has come as a direct result of the AI race.
In an interview with The Hill last year, Google’s executive chairman Eric Schmidt said that AI is “a game changer.”
“It’s not going to be the dominant force in our lives, but it is going to transform the way we live and work,” Schmidt said.