3 Ways Tomorrow’s AI Will Differ From Today’s Chatbots
After interviewing the makers of ChatGPT earlier this week, I’m left pondering: How exactly would I explain my job to an ancient hunter-gatherer? I’m thinking: “I put words and pictures into a machine that makes a clickety-clack sound and then those words and pictures can be seen by others with clickety-clack machines." How would you explain your job? During my interview with OpenAI Chief Executive Sam Altman and Chief Technology Officer Mira Murati at WSJ Tech Live—which you can watch above or here—Altman said AI will change work faster than other major technological revolutions did. And jobs will change so much they could be unrecognizable to those of us employed today. My columns are typically focused on tech’s here-and-now, but this conversation was about AI’s rapid evolution in the coming years.
Here are three things I took away from the conversation that surprised me: AI will require less data to become as smart as humans. One of the goals of OpenAI is to create AGI—or artificial general intelligence. Murati defined it as “a system that can generalize across many domains that would be equivalent to human work" and that it will produce “a lot of productivity and economic value." When I asked Altman why this is the goal, he said it will be “the best tool humanity has yet created." With it, we will be able to solve unsolvable problems and create “incredible things," he added.
Altman and Murati wouldn’t give me a concrete timing on when AGI will arrive. Maybe within the next 10 years? But they did say that to get there, they won’t need an internet’s worth of training data. What matters more will be the quality and value of the data it can access as it gets better at reasoning.
Read more on livemint.com
