With the arrival of exceptionally smart chatbots like ChatGPT and Bard, the debate over AGI has shifted from “if” to “when.”
Artificial general intelligence (AGI) is the talk of Silicon Valley these days. With extremely capable chatbots like ChatGPT hitting the scene, tech CEOs are more jazzed than ever about AGI becoming a reality soon.
But what is AGI anyway? As it stands, there’s no textbook definition. But basically, it’s AI that can handle pretty much any mental task a human can, or even excel at it. This includes planning, learning new skills, making judgments, and using past knowledge to make decisions or get more accurate.
As for when will this advanced AI actually emerge, big tech CEOs have plenty of takes. With ChatGPT and Google’s Bard causing a stir, the debate has shifted from “is AGI possible?” to “when will it get here?” Let’s see what the experts predict.
Elon Musk (Founder of xAI)
Back in July 2023, Elon Musk held a 90-minute Twitter Spaces session for over 30,000 listeners when he shared details about his latest venture, xAI. During the chat session, Musk made it clear he believes AGI will happen – and soon.
“I think we can make something better than DeepMind or OpenAI,” Musk claimed, although he didn’t explain how exactly. But Musk did lay out a specific timeline for when he thinks usable AGI will hit the scene – around 2029.
Shane Legg (Co-founder and chief AGI scientist at DeepMind Technologies)
Over 10 years ago, Shane Legg predicted AI would have a 50/50 chance of matching human smarts by 2028. And he’s still betting on that forecast.
In a recent podcast, Legg said he thinks researchers have equal odds of pulling off AGI in the next few years. But he noted some caveats. First, defining AGI depends on nailing down human intelligence – and our thinking is complicated. “You’ll never have a complete set of everything people can do,” Legg pointed out. Still, if AI aces enough tests of human smarts, that’s AGI in his book.
Legg’s other big caveat is that AI training has to be scaled up significantly. Quizzed about where we stand now on the road to AGI, Legg noted that we’ve got the computing power. But the “first unlocking step” is training models on more data than a human sees in a lifetime.
Geoffrey Hinton (Former VP at Google, “Godfather of deep learning”)
AI concerns Geoffrey Hinton so much that back in May this year, he left Google to warn people of its dangers. After exiting, Hinton made a new prediction on when AI will beat human intelligence. Ominously, the “godfather of deep learning” drastically moved up his original 30-50 year forecast.
“I now predict five to 20 years but without much confidence,” he tweeted. “We live in very uncertain times. It’s possible that I am totally wrong about digital intelligence overtaking us. Nobody really knows which is why we should worry now.”
Yann LeCun (VP and chief AI scientist at Meta, Turing Award winner)
Yann LeCun won the Turing Award in 2018 with Geoffrey Hinton and Yoshua Bengio for their groundbreaking AI contributions. But unlike his fellow pioneers, LeCun doesn’t see AI as a threat to humanity.
“Will AI take over the world? No, this is a projection of human nature on machines,” he said at a June 2022 press event. Keeping AI research “under lock and key” would be a huge mistake in LeCun’s view.
LeCun agrees AI will eventually outsmart humans. But he says we still lack key concepts needed to reach that level, which could take many years or decades. He also said in a response to a question by BBC News that AI progress would be gradual, perhaps reaching rat brain-level intelligence. He stressed such an AI would still run in a controlled data center and could be shut off if needed. LeCun added that if it seems unsafe, the logical move is to not build it at all.
Sam Altman (CEO of OpenAI)
According to Sam Altman, two things will matter most in coming decades for improving life – AGI and affordable clean energy. He believes transformative AGI is on the horizon, calling it potentially humanity’s most powerful tool ever.
“We’ll be able to express ourselves in new creative ways. We will make incredible things for each other, for ourselves, for the world, for this unfolding human story,” envisions Altman. Despite inevitable growing pains, he sees a future where AGI’s benefits are so clear current concerns will seem overblown. Altman pictures a society enriched by amazing innovations that add to humanity’s epic story.
When asked about expected AGI timelines, Altman said it could emerge in the next decade.
Demis Hassabis (CEO of DeepMind Technologies)
Demis Hassabis recently predicted AI matching human smarts in “the next few years” or “maybe within a decade.” Hassabis co-founded DeepMind, creators of the AlphaGo AI that defeated top human Go champs.
In a Wall Street Journal festival interview, Hassabis told Chris Mims he thinks human-level AI is arriving soon: “The progress in the last few years has been pretty incredible. I don’t see any reason why that progress is going to slow down. I think it may even accelerate. So I think we could be just a few years, maybe within a decade away.”
So there you have it – from warnings to optimism, it seems the tech experts agree game-changing AI looms on the horizon. Even as there are many others sceptical if we’ll even see superintelligent bots in our lifetimes. For now, it seems like all we can do is wait and see what the future has in store.
Source:indianexpress.com