The Beatle who wasn’t—and the quiet leadership crisis AI creates

In 1965, Paul McCartney walked into Abbey Road Studios with a simple acoustic number called “Yesterday” and asked producer George Martin what it needed. Martin’s suggestion was startling: a string quartet, on a Beatles rock record, at a time when nothing like that had been tried. McCartney resisted; Martin persisted. The result became one of the most recorded songs in history—not because Martin was a better songwriter, but because he heard a possibility the songwriter couldn’t.

Sixty years later, that story has become an unlikely parable for the AI era. As artificial intelligence floods organisations with instantly generated business plans, market analyses, legal briefs, and code, the scarce resource is no longer answers. The competitive advantage has shifted to judgment—the ability to ask better questions, to reframe problems, and to hear the creative potential buried in a raw idea.

The author, a leadership advisor, argues that most companies still treat AI as a tool for efficiency gains, missing its deeper value. When every team can produce the same analysis in minutes, the differentiator is not faster execution but the willingness to challenge the assumptions baked into the request. Like George Martin, the leader’s job is not to out-produce the machine but to listen for what it cannot yet hear.

What a 1965 recording session reveals about the end of the answer era

The producer's edge: hearing what the talent couldn't

George Martin was not deaf to the Beatles’ genius; he amplified it by bringing skills they lacked—classical arrangement, tape manipulation, and the honesty to tell the band when a song wasn’t working. Critically, he didn’t force the group into his own classical comfort zone. He pushed “Yesterday” toward a string quartet because that is what the song needed, not because it was what he knew best. For leaders today, the instruction is to use AI not to replicate yesterday’s expertise but to uncover possibilities the team’s own maps don’t yet show.

Why the efficiency playbook fails when everyone has the same tools

The piece identifies a trap: most organisations treat AI as an optimisation engine—summarising meetings, drafting reports, automating customer service. While those gains are real, they are fleeting. Efficiency stops being a competitive advantage the moment every rival deploys the same large-language models. The bottleneck shifts to the assumptions, habits, and decision-making structures built for a world where intelligence was expensive and answers were slow to arrive. Companies that simply bolt AI onto existing hierarchies will end up doing the wrong things faster.

The author proposes three levels of AI leadership. Level one is optimisation, which is essential but insufficient. Level two is simulation: using AI defensively to war-game plans, stress-test strategies, and hunt for blind spots before a competitor does. Level three—the creative level—is where the most value lies: asking what might be created together that neither human nor machine could achieve alone. That is the George Martin level, and it demands leaders who prize questions over certainties.

Underneath all three levels is a more fundamental shift. Intelligence has been democratised; judgment has not. The machine can flood a meeting with probable paths, but only a human can decide when the map itself is wrong. The leadership challenge, then, is to build organisations that reward the act of reframing problems, not just solving them.

Three ways to lead with AI that actually matter

  • Before your next strategy offsite, ask an AI model to generate the recommendation your team would likely propose. If it does the job in minutes, redirect the offsite to stress-testing why that recommendation might be flawed—not to building it from scratch.
  • Pick one deeply held assumption in your industry (e.g., “customers will never pay for X” or “regulation will stay constant”) and have the AI simulate three plausible scenarios in which that assumption collapses within two years. Use the output to pressure-test your three-year plan.
  • Design a monthly “red team” ritual: feed a key plan to the AI with the instruction to play the role of an aggressive competitor and dismantle the plan point by point. Share the results with the leadership team before any investment decision is finalised.
  • When evaluating new AI use cases, separate the efficiency projects (likely to be copied by competitors) from the creative experiments that could redefine a market. Fund the latter with the savings from the former.