For the past three years, “learn prompt engineering” has been the loudest career advice in every boardroom, LinkedIn feed, and corporate training deck. But PwC’s recent AI blunder — where the consulting giant’s reports were found riddled with fabricated citations and hallucinated research — has flipped that advice on its head. According to AI thought leader Jaspreet Bindra, founder of AI&Beyond and former Chief Digital Officer of the Mahindra Group, the most valuable workplace skill today is no longer the ability to prompt AI brilliantly. It is AI verification skills: the discipline of checking, questioning, and validating what AI produces before it goes out into the world.
The PwC Blunder That Changed the Conversation
The trigger for this shift was embarrassing precisely because of who it happened to. PwC — a firm that sells “responsible AI” advisory services to global corporations — published thought leadership reports containing fake academic citations, non-existent studies, and unverified data, apparently generated by AI and pasted into official documents without adequate human review. The errors were flagged by AI-detection researchers at GPTZero and later verified by the Financial Times. EY and KPMG faced similar embarrassments, retracting AI-tainted papers of their own.
The lesson, as Bindra frames it, is brutal in its simplicity: the failure was never the AI’s. Large language models are known to hallucinate — they generate fluent, confident, plausible-sounding text with no guarantee of factual accuracy. The failure was human. Nobody verified. The world’s most sophisticated professional services firms had mastered the input side of AI — the prompting — while completely neglecting the output side: verification. That gap is exactly where AI verification skills have become the defining differentiator of the modern professional.
Jaspreet Bindra’s POV: From “Homo Promptus” to the Human Verification Layer
Jaspreet Bindra has long been one of India’s most influential voices on AI and the future of work. Author of The Tech Whisperer and Winning with AI, with a Master’s in AI, Ethics and Society from the University of Cambridge, Bindra popularised the idea of “Homo Promptus” — the new professional whose superpower is knowing how to talk to machines. Which makes his current position all the more striking: even the man who championed prompting now argues that prompting alone is dangerously incomplete.
Bindra’s core arguments can be summed up in four points.
First, prompting has been commoditised; judgment has not. As AI models get better at understanding plain, conversational language, the craft of engineering the “perfect prompt” matters less and less. The models increasingly meet users halfway. What machines cannot do — and what the PwC episode proved — is take accountability for their own output. That remains an irreducibly human job, and it is where the scarce, valuable skill now lives.
Second, AI is a brilliant intern, not an oracle. Bindra has consistently described generative AI as a tireless, talented assistant that must be supervised like one. You would never send an intern’s first draft to a client unchecked. Treating AI output as finished work — as PwC’s teams effectively did — is not an AI failure but a management failure. The professional’s role shifts from creator to editor-in-chief: reviewing, fact-checking, and signing off.
Third, critical thinking is the real AI skill. In his talks and writing on the future of work, Bindra lists the abilities that will matter most in an AI-first workplace: yes, knowing how to use AI tools, but crucially, the critical thinking needed to spot hallucinations and biases in AI outputs. AI verification skills are, in his framing, applied critical thinking — asking “How do I know this is true?”, tracing citations back to sources, cross-checking data, and being professionally sceptical of anything that reads too smoothly.
Fourth, trust is the currency, and humans are its guarantors. Bindra’s larger point goes beyond skills to ethics and accountability. Consulting firms, law firms, banks, and media houses sell one thing above all: trust. When AI-generated errors slip into client-facing work, the damage is not a single bad report — it is the erosion of the institution’s credibility. Ethical AI use, he argues, is not a compliance checkbox but a necessary safeguard, and the “last mile” of any AI workflow must be a human being who verifies and stakes their name on the result.
What This Means for Your Career
The practical implication of Bindra’s view is a rebalancing of what professionals should learn. Prompt engineering got you into the AI conversation; AI verification skills will keep you employed in it. That means building habits such as source-checking every citation an AI produces, validating numbers against original data, running AI claims past domain experts, and maintaining a clear human sign-off step in every AI-assisted workflow.
It also means organisations must redesign processes, not just buy tools. The PwC blunder happened inside a firm with world-class talent and a billion-dollar AI investment — proof that access to AI without a verification culture is a liability, not an advantage. Companies that reward speed of AI output without rewarding rigour of AI review are, in effect, automating their own reputational risk.
The Bottom Line
Jaspreet Bindra’s message lands with particular force because it comes from an AI optimist, not a sceptic. He is not telling professionals to abandon AI — he is telling them the game has moved. In the first wave of generative AI, value went to those who could get the best answers out of the machine. In this second wave, value goes to those who can tell whether those answers are actually true. After PwC’s AI blunder, AI verification skills have overtaken prompt engineering as the most valuable workplace skill — because in an age when machines can write anything, the humans who can verify everything become indispensable.