Zoho co-founder Sridhar Vembu believes AI creativity lies outside the training distribution: Here's why

Zoho co-founder and chief scientist Sridhar Vembu has weighed in on the ongoing debate around artificial intelligence and creativity, arguing that true creative work occurs “out of the training distribution”, a realm where large language models (LLMs) often struggle.

Zoho’s Sridhar Vembu on AI and Creativity

In a post shared on X on Saturday, Vembu said that while AI systems such as chess or Go engines can produce creative moves, they rely on a fundamentally different approach compared to modern LLMs.

“True creative work is ‘out of the training distribution’ work,” Vembu wrote. “Chess or Go engines do come up with creative moves. The foundational approach they use, Monte Carlo Tree Search, is different from how LLMs work and that may explain why LLMs don’t do too well ‘out of their training distribution’.”

He further explained that games offer a structured environment with precise rules distinguishing valid from invalid moves, a stark contrast to the complexities of the real world.

“Games have precise rules of the game, valid vs invalid moves etc. The real world is much more messy,” he noted, adding that software code shares some characteristics with games. “Software code has some of the character of games and that is one area that can look to techniques from game engines.”

Vembu’s remarks highlight a growing discussion within the tech community about the limitations of current AI models, particularly in performing tasks that require genuine innovation or reasoning beyond their training data.

Vembu Faces Backlash Over Vaccine Remarks

On Tuesday, the Zoho co-founder sparked controversy after sharing a research paper that claimed vaccination is the “dominant risk factor” for autism. Posting on X, Vembu urged parents to “take this analysis seriously,” adding, “There is increasing evidence that we are giving way too many vaccines to very young children. This is spreading in India too, and we are seeing a rapid increase in autism.”

His remarks drew sharp criticism from medical experts. Dr Philips, an award-winning hepatologist recognised for countering medical misinformation, condemned the claims, pointing out the paper’s lack of scientific credibility. “The so-called ‘authors’ are known anti-vaccine activists funded by an anti-vaccine organisation,” he wrote. “The study was published on their own website and has not undergone any peer review or independent scientific scrutiny.”