Are We Having the Wrong Conversation on AI and Jobs?

Key Takeaways

• Anthropic study finds no immediate rise in unemployment among exposed workers.
• Human tacit knowledge and practical wisdom limit LLMs’ task automation.
• AI adoption resembles adding a dynamo to steam factories, not redesigning workflows.
• Productivity gains risk favoring capital owners, echoing the historical Engels Pause.

Pulse Analysis

The buzz around AI replacing white‑collar jobs has outpaced the evidence. Anthropic’s latest labour‑market study plotted the share of tasks large language models could perform against actual AI usage, revealing a massive capability‑adoption gap. Early signals, such as a modest slowdown in hiring younger workers for highly exposed roles, suggest that the feared “white‑collar bloodbath” is not materialising. This disconnect underscores a broader lesson: raw technological potential does not instantly translate into labour displacement, and policymakers should temper alarmist forecasts with real‑world adoption data.

A deeper obstacle lies in the tacit dimension of work—knowledge that resides in experience, intuition, and on‑the‑fly judgment. Truck drivers, consultants, and account directors illustrate how seemingly routine tasks embed complex decision‑making that LLMs cannot codify. As Ian Leslie and Michael Polanyi have noted, practical wisdom (phronesis) is essential for navigating ambiguous situations, making AI a complement rather than a substitute. Historical analogues, from the introduction of ATMs to the shift from steam to electricity, show that technology often first augments existing workflows before a paradigm shift prompts a wholesale redesign of processes and organisational structures.

The most pressing risk, however, is not job loss but the distribution of AI‑driven productivity gains. The Engels Pause during the early Industrial Revolution saw output soar while wages stagnated, enriching capital owners at workers’ expense. Early signs of a similar pattern appear today, with firms leveraging AI to boost efficiency without proportionate wage growth. To avoid a modern‑day Engels Pause, businesses and regulators must focus on redesigning work systems, fostering new skill sets, and ensuring that the economic upside of AI is shared broadly across the workforce.