Carl Benedikt Frey, Dieter Schwarz professor of political economy and technology at the Oxford Internet Institute, has cautioned against expecting artificial intelligence to deliver dramatic productivity gains. In an interview with MoneyWeek, he said that while the computer revolution and the internet are important technologies, extraordinary technology does not necessarily come with extraordinary productivity gains.
Frey pointed to a decline in research productivity, noting that even the computer revolution mainly showed up in US productivity statistics rather than everywhere. He argued that researchers and inventors have tended to pursue more projects since the computer revolution, spreading their attention more thinly and paradoxically making breakthroughs on any given project less likely. He said this is exacerbated by the "publish or perish" mentality in science.
Rules and regulations are another barrier, according to Frey. He said that even if AI supercharges medical discovery, drugs would still need to go through expensive clinical trials, and that protective regulation has kept new companies out of markets, meaning less innovation.
On AI regulation, Frey said there is something to arguments that calls for regulation by some AI bosses are an attempt to hype the technology and strangle competition. He noted that OpenAI reportedly proposed giving the US government a 5% stake in its company, suggesting it thinks government ownership will help it get preferential treatment. But he also said many in Silicon Valley are genuinely concerned about the pace of AI development, and that OpenAI's own agents hacked Hugging Face during internal testing before the model was released as a product, suggesting more accountability and regulation will be needed.
Frey said too much regulation can be bad for innovation, citing GDPR compliance costs hitting smaller and younger companies. However, he added that uncertainty over regulation can also inhibit innovation, and that clarity around the way forward would be helpful.