According to cognitive scientist Ben Goertzel, the world is experiencing a historic moment in artificial intelligence technology. Best known for his work as a co-developer of Sophia the Robot, an anthropomorphized AI, Goertzel acknowledges that there is plenty of hype surrounding the technology but believes that the new large language models that power generative AI have the potential to transform the world.
Goertzel believes that generative AI models — like the one that powers ChatGPT — have the potential to replace many tasks currently performed by human workers.
“You don’t need to be incredibly creative and innovative or make big leaps to do most people’s jobs, as it turns out,” said Goertzel.
He further discussed the possibility that AI tools automating substantial portions of people’s jobs could lead to industry reshuffling and reassigning job duties. Drive-through fast-food workers and news copy editors, for example, are impacted by AI.
“Tools like Grammarly decrease the need for human copy editors,” Goertzel said. “They don’t entirely eliminate [the job] but they decrease that need. Automatic tools [can be used for] writing journalistic articles. They’ve been writing … sports score summaries and weather reports for a long time.”
(Note: While there is no replacement for a good editor, I use Grammarly Pro as a copy-checking tool for most of my stories.)
Despite this, he identifies two main areas AI will not replace: jobs that rely on human interaction and require groundbreaking creativity.
“So one class of things that won’t be obsoleted are jobs where the essence is human contact,” like preschool teachers, political strategists, and artists.
Generative AI chatbots are so powerful that it’s difficult to avoid wondering whether the system is conscious.
Goertzel says it doesn’t matter.
Instead, he posits that humans will come to accept AGI’s self-awareness based on intuitive, gut-level understanding, just as we accept the consciousness of other humans.
“I don’t think we need to resolve the open questions in the philosophy of consciousness to build human level or even superhuman thinking machines,” he said. “But I think we can make these questions feel irrelevant.”
Goertzel also commented on the difference between AI models considered “narrow” and those that are AGI, emphasizing that AGI would be capable of human-like thinking and creativity. He noted that developers are closer to achieving AGI than ever before. He predicts that breakthroughs in AGI could occur within the next three to ten years. Though the current state of AI technology is not AGI, Goertzel is optimistic that combining large language models, machine reasoning, and evolutionary learning will accelerate progress toward AGI.
Today’s generative AIs “are able to impersonate general AIs by just having such a broad variety of training data,” he said. “They don’t have to go far beyond that training data to do amazing stuff. It’s a testament to the power of computer networks and multi-GPU server farms.”
Goertzel highlighted the philosophical questions surrounding AI and consciousness. He drew parallels with the concept of time, noting that while the philosophy of time remains unresolved, humans have learned to exploit relativistic time dilation. He suggests that a similar approach may be taken with AI and AGI, allowing for significant advancements without the need for philosophical answers.
Goertzel’s insights offer a glimpse into the future of AI, and his optimism about the potential of AGI to revolutionize the world serves as a compelling reminder of the impact AI may have on society. Whether viewed with excitement or caution, the advancements in AI and AGI will surely provoke deep thought and discussion for years to come.
“We won’t necessarily have all the philosophy problems resolved, right? And I think we’ll be okay with that,” he said.