The AI Productivity Paradox: A Slow-Motion Train Wreck?

Top economist Torsten Slok has sounded an alarm about the yawning gap between artificial intelligence-enhanced productivity and actual returns on investment. This disparity is starkly evident in the data, which shows that profit margins for the S&P 500 have stagnated at around 12%, while those of top tech companies like Apple, Amazon, Microsoft, Alphabet, Facebook, Tesla, and Nvidia have skyrocketed from 15% to 25% over the past three years.

The disparity is not just a matter of regulatory hurdles or workflow integration issues. Rather, it’s a fundamental problem that stems from the pressure to deliver on AI’s promise of transformative productivity gains. Companies will slow their AI spending if they don’t see returns quickly, and Slok’s warning sign suggests that AI implementation might be a bumpier road than expected.

Ford’s recent admission that its mass automation efforts have hit a snag is telling. The automaker has deployed AI vision systems across 33 global plants but recognized the technology was not as effective without human oversight. Charles Poon, Ford’s vice president of vehicle hardware engineering, candidly acknowledged that “artificial intelligence is a fantastic tool, but it’s only as good as the information you use to train it.”

The phenomenon of “AI shame” – where companies feel pressured to adopt emerging technologies without articulating tangible use cases or goals – is another factor driving this productivity paradox. Deploying AI just for the sake of it may actually reduce productivity gains, as Boston Consulting Group found. Tokenmaxxing, a practice in which tech companies incentivize AI use through internal leaderboards, has led to workers using the technology for the sake of it, driving up costs.

Slok’s warning is that a painful repricing of markets is possible if the gap between earnings expectations and actual returns on investment continues to grow. This sobering reminder underscores that AI is not a panacea for productivity woes – at least not yet. As Peter Cappelli, a professor of management at the University of Pennsylvania’s Wharton School, noted in a case study on Ricoh, people are greatly underestimating “just how much work is involved” in realizing productivity and ROI gains.

In reality, AI has delivered more hype than substance so far. We’ve been sold a bill of goods that AI will revolutionize productivity, but the evidence suggests otherwise. As we hurtle towards an AI-dominated future, it’s essential to separate hype from reality – and recognize that even the most promising technologies require time, effort, and practical implementation to deliver on their promise.

The clock is ticking, indeed. But if we’re not careful, we might just find ourselves stuck in a slow-motion train wreck, with AI productivity gains stalled at the station.