Databricks CEO says we already have artificial general intelligence. Silicon Valley simply refuses to acknowledge this.
Ali Ghodsi said at the Goldman Sachs Communicopia + Technology conference in September that AI chatbots already meet the definition of AGI – AI that can reason as a person – this is what researchers used ten years ago.
“Everyone would have said yes, but we kept changing the goalposts,” Ghodsey said in a discussion posted Tuesday.
“Now that we've achieved this, let's come up with something even bigger,” he added, meaning strive for superintelligence – AI that can reason much smarter than humans.
Ghodsey, who has a PhD in computer science, said the industry's goal of creating superintelligence is a far cry from today's methods. He also said the focus on superintelligence is “misdirected,” adding that building systems that can outsmart the world's brightest minds is not what companies actually want.
AGI “has everything we need to automate and create agents,” he said. “We just have to do the boring work,” he added.
San Francisco-based Databricks raised $1 billion in September, valuing the company at more than $100 billion.
Ghodsey said at the conference that the era of giant leaps from artificial intelligence models has slowed. The scaling laws that have driven AI progress over the last few years have clearly stalled, and new systems like OpenAI's GPT-5 and Anthropic's Claude 4 aren't making significant improvements.
“It’s getting harder and harder to capitalize on the next pre-trained giant model,” he said.
Superintelligence debate
Ghodsey's comments come amid industry disagreement over whether artificial superintelligence is even desirable.
Microsoft General Manager of Artificial Intelligence said in an episode of the “Silicon Valley Girl Podcast” published Saturday that artificial superintelligence should be treated as an “anti-goal.”
Superintelligence “doesn't seem like a positive vision for the future,” Mustafa Suleiman said. “It would be very difficult to contain something like that or bring it into line with our values.”
Suleiman, who co-founded DeepMind before joining Microsoft, said his team is instead aiming for what he calls “humanistic superintelligence”—one that is based on human interests and values.
Other technology leaders are determined to achieve superintelligence.
OpenAI CEO Sam Altman said earlier this year that the company is moving toward superintelligence without stopping at AGI.
“Superintelligent tools have the potential to dramatically accelerate scientific discovery and innovation far beyond what we can do on our own, and in turn greatly increase abundance and prosperity,” Altman said in January.
Altman said in an interview in September that he would be very surprised if the industry has not reached supermind by 2030. Altman has long described AGI as OpenAI's core mission.
Google DeepMind co-founder Demis Hassabis proposed a similar graph. In April, he said AGI could emerge “in the next five to ten years,” describing a future in which artificial intelligence systems understand the world “in a very subtle and deep way” and are woven into everyday life.
