India shouldn’t ‘bet everything on LLMs,’ says Oxford’s Carl Benedikt Frey at Davos

At the World Economic Forum in Davos, Carl Benedikt Frey, a professor at Oxford University, urged India to avoid putting all its resources into large language models (LLMs). Instead, he emphasized the importance of embracing open-source technologies. Frey, who serves as the Dieter Schwarz Associate Professor of AI & Work at the Oxford Internet Institute, expressed concerns over whether the current AI landscape will be sustainable in the long run. Frey believes that India possesses a unique opportunity to establish itself as a global AI powerhouse by expanding its research efforts beyond popular trends like LLMs. He advocates for a strategy that includes leveraging open-source models in areas where India already excels, while also investing in diverse research avenues such as small language models and symbolic AI. "If India charts its own course and makes informed decisions, it has the potential to lead in the AI sector," Frey stated. However, he also raised alarms about the precarious nature of the AI market. He warned of a 50% chance that the AI bubble could burst by next year, noting that there has been a decline in AI adoption since last summer. Frey pointed out that current market valuations lack the necessary growth in use cases to remain viable. Addressing concerns about AI-induced job losses, Frey offered a counterintuitive viewpoint. He suggested that AI is actually lowering barriers to entry in knowledge work, which could be advantageous for India’s economy, especially given the existing wage disparities. "If AI helps to reduce the productivity gap, we might see more jobs moving to India," he predicted. He also highlighted the rising hiring rates in India’s service sector, contrasting this with declines in employment in Europe and North America.