“It’s not an overreach.”
Shark Thoughts
Stressed about an AI-wrought apocalypse? Look no further than one Mark Cuban, extremely rich investor of “Shark Tank” fame, to validate your AI anxiety.
Asked by The New York Post whether recent AI doomsday warnings from the likes of OpenAI CEO Sam Altman and since-defected OpenAI co-founder Elon Musk were too extreme, Cuban responded with what reads as a version of “nah.”
“It’s not an overreach,” Cuban told the Post. “It’s a request for people to pay attention.”
“AI as it is today is not an existential threat,” the billionaire added. “But in the future, say 15 to 20 years, there will be simulations of threat creation and responses for things we haven’t even considered to this point.”
In other words, according to Cuban, folks like Altman and Musk are rightfully sounding the alarm — not necessarily about what’s happening in the present, but about looming unknowns in our not-so-distant future.
Case to be Made
The guy’s got a point. Existing AI presents serious concerns, the threat of mass-produced misinformation being a particularly pressing one. But at-scale misinformation, while obviously a bad and harmful thing, isn’t necessarily threat-level doomsday.
And importantly, something like synthetic content-inflicted information scrambling is a fairly predictable outcome of this kind of tech. The existential dangers of AI systems, as Cuban notes, might really lie in their unpredictability.
AI, after all, is still in many ways a black box. Most of the time, researchers don’t really know what’s going on inside a model’s decision-making process, hence the reason why the AI industry’s most popular meme is of a horrifying and mysterious monster masked by a shiny little smiley face. It’s impossible to soundly predict future behaviors of a system we don’t fully understand — especially considering that we might unknowingly be reinforcing some bad, black box-hidden behaviors.
Century Club
If it makes you feel any better, Cuban did add that he doesn’t think anything too “sci-fi” would happen for another century or so.
“It’s not inconceivable that we will record everything we see, say, write and do in our lives, store it in an AI, effectively being a version of our own brains,” Cuban, in true billionaires-who-want-to-live-forever fashion, told the Post. “Combine that with the replication of ourselves and the AI can live on for us after we are gone.
“Insane,” he added, “but definitely on the 100-year knowledge curve of AI and precision biology that is beginning now.”
More on AI doomsday: OpenAI CEO Hopeful World Leaders Will Save Us From AI He’s Building