Artificial intelligence tools excel at automating routine tasks but fail to deliver the scientific breakthroughs needed for sustained economic growth, according to Carl Benedikt Frey from Oxford University. Writing in the Financial Times, Frey argues that despite decades of technological advancement, labor productivity growth has declined from 2 percent annually in the 1990s to 0.8 percent in recent years. Large language models tend toward statistical consensus rather than revolutionary thinking, he notes. A survey of 7,000 knowledge workers showed AI reduced email tasks by 31 percent but left collaborative work unchanged. Historical examples demonstrate that economic miracles stem from discovery, not just increased efficiency. Frey suggests AI could still boost productivity if institutions reward originality over volume and support riskier research endeavors.