Verification is where AI's economic value gets realised

Summary

An MIT Sloan research article argues that AI's productivity gains do not translate to economic value without reliable verification of its outputs. As AI systems become more capable, the gap widens between what they can produce and what humans can audit, and human capacity to verify becomes the bottleneck. The piece lays out the risks of skipping verification or using AI to verify AI, traces how verification skill erosion creates a talent gap (the "missing junior loop"), and offers implications for firms, workers, and policymakers. Full article.

Key Claims

• AI's productivity edge collapses without verification capacity: code generation jumped from 4.4% to 71.7% accuracy on SWE-bench in a year, but human verification bandwidth remains scarce and fixed.
• Verification is now a value-creation function, separate from compliance; firms that understand and underwrite AI risks will profit, while those treating outputs as reliable without checking accumulate "technical debt."
• Using AI to verify AI is a "tempting shortcut" but fails when both systems share the same assumptions and errors, creating false confidence rather than real assurance.
• As AI displaces entry-level roles, it erodes the junior-level training ground where verification skills in senior staff develop—the "missing junior loop"—shrinking the pool of experienced auditors.
• Firms must scale automation only as fast as it can be trusted; competitive pressure to deploy unverified systems creates hidden risk and unobserved failures that can eventually cascade.

Quotes

• "AI makes it cheap to produce work, but not to judge whether that work is any good."
• "The companies that understand the risks and can underwrite them will be the ones that profit."
• "If we do not invest in verification, we're accumulating hidden risk. It is technical debt accumulating behind the scenes, and, at some point, it'll come due."
• "The ladder is breaking."