The AI Productivity Paradox

An NBER study analyzing responses from 6,000 CEOs, CFOs, and executives in the US, UK, Germany, and Australia found that about 90% of companies said AI had no impact on employment or productivity over the past three years.

While about 2/3 actually used AI, the average usage time was only 1.5 hours per week, and 25% did not use AI at all in their work.

Nevertheless, executives expected that over the next three years AI would bring

• a 1.4% increase in productivity
• a 0.8% increase in output
• a 0.7% decrease in employment while a separate survey of individual workers showed employment expected to rise by 0.5%.

FT's analysis notes that 374 S&P 500 companies mentioned AI in earnings calls, but macro indicators have yet to reflect a clear effect. Some studies show productivity gains, but results from the Fed, MIT, Stanford, and others suggest the effects are small or dispersed.

The key issue is the time lag between AI investment and actual productivity gains. According to the article, AI's value depends less on adoption itself than on how it is applied to work, and we are now in a phase where Solow's old productivity paradox is resurfacing.