Hoffman, who helped fund OpenAI in its early days and now co-founded the AI drug-discovery startup, Manas AI, believes today’s most advanced AI models can provide insights that doctors would be unwise to overlook.
“These AI systems, even though many of them are not specifically trained for medicine, have ingested trillion-plus words of information,” he said at WIRED Health in London earlier this spring. “As a second opinion, it is bringing superpowers that no human being has.”
AI as a second opinion
Hoffman wasn’t suggesting that AI replace doctors, but rather that it be used as part of the decision-making process. He said he regularly uses AI as a second opinion on his own health concerns and expects his concierge physicians to do the same.
“You could very well go, ‘No, I think you’re wrong, I think it’s this,’” Hoffman said. “But if you’re not using this as a second opinion, you’re making a mistake, both as a doctor and as a patient.”
Not everyone agrees that AI is ready to play a larger role in healthcare. A recent study led by researchers at the University of Oxford found that people using large language models to assess medical scenarios made decisions that were no better than those relying on traditional online searches or their own judgment.
“Despite all the hype, AI just isn’t ready to take on the role of the physician,” said Dr. Rebecca Payne, a GP and lead medical practitioner on the study. “Patients need to be aware that asking a large language model about their symptoms can be dangerous, giving wrong diagnoses and failing to recognise when urgent help is needed.”
While experts continue to debate its limitations, AI is already finding its way into doctors’ offices and hospitals. A recent Philips Future Health Index survey found that 27% of clinicians said AI helped them identify potential medical errors at least three times in the previous three months, while 36% said it allowed them to see more patients each week.
Still, the survey suggests the industry’s embrace of AI remains a work in progress. Despite growing interest in the technology, 77% of clinicians said AI training was unavailable, limited or inconsistent.
