Artificial general intelligence is still years away and far from the breakthroughs some tech leaders predict, former Tesla Inc. (NASDAQ:TSLA) senior director of AI and OpenAI founding member Andrej Karpathy said.
"We're at this intermediate stage," he told host Dwarkesh Patel on the "Dwarkesh Podcast" in October. "The models are amazing. They still need a lot of work."
Hype Meets Reality
Karpathy said the rapid growth of large language models has created unrealistic expectations. "Overall, the models are not there," he said. "I feel like the industry is making too big of a jump and is trying to pretend like this is amazing, and it's not."
Don't Miss:
- Buffett's Secret to Wealth? Private Real Estate—Get Institutional Access Yourself
- Wall Street's $12B Real Estate Manager Is Opening Its Doors to Individual Investors — Without the Crowdfunding Middlemen
He said some companies promote "AI agents" — systems that write code, search the web, or complete office tasks — as if they are truly autonomous. He told Patel such systems still lack reliability and genuine understanding.
His view contrasts sharply with that of OpenAI CEO Sam Altman, who told German newspaper Die Welt in Berlin in September that he would be "very surprised" if by 2030 the world didn't have "extraordinarily capable models that do things that we ourselves cannot do."
Why He Believes It Will Take Ten Years
Karpathy said in the podcast AI still struggles with structured reasoning, long-term memory, and safety. He said that many demonstrations show narrow skills rather than true general intelligence.
"I'm very unimpressed by demos," Karpathy said. "Whenever I see demos of anything, I'm extremely unimpressed by that." He said that polished demonstrations rarely capture real-world complexity.
Trending: Missed Nvidia and Tesla? RAD Intel Could Be the Next AI Powerhouse — Just $0.81 a Share
He called his outlook "five to 10 times pessimistic" compared with public forecasts but said a decade-long path should still be considered optimistic. "Ten years should otherwise be a very bullish timeline for AGI," he wrote later on X.
A Reality Check For Silicon Valley
The contrast in timelines underscores how divided the tech world remains over AI's progress. Tesla CEO Elon Musk wrote on X last month that his company's upcoming model, Grok 5, "will be AGI or something indistinguishable from AGI," adding that there's a "10% chance of achieving AGI and rising."
Meanwhile, Microsoft Corp. (NASDAQ:MSFT) CEO Satya Nadella told developers at GitHub Universe 2025 in San Francisco last month that "a new toolchain is evolving and is being born right now," as AI agents transform how software is built.
See Also: If there was a new fund backed by Jeff Bezos offering a 7-9% target yield with monthly dividends would you invest in it?
Karpathy offered a more cautious perspective, saying current AI systems remain far from general intelligence. "We have some very early agents … but I still feel there's so much work to be done," he said on the podcast.
Still, he remains confident the hurdles can be overcome. "I feel like the problems are tractable, they're surmountable," he said. "But they're still difficult."
Read Next: 7 Million Gamers Already Trust Gameflip With Their Digital Assets — Now You Can Own a Stake in the Platform
Image: Shutterstock
