Never turn to AI for thought leadership

It was supposed to be a landmark report – a set of recommendations from Deloitte designed to reform a controversial aspect of Australia’s welfare system. Instead, it turned into a farce. After Deloitte found multiple issues with the automated system used to issue penalties to job seekers if they didn’t meet certain conditions set by the government, the report itself was soon criticised by an academic at the University of Sydney for including fake citations. 

After an internal review, Deloitte updated the document, revising the citations and including a disclaimer that it had partially been written using GPT-4o. Despite being forced to refund part of the £215,000 fee it originally charged the Australian government for the privilege of investigating its welfare system, the consultancy said it stood by its conclusions. “The updates made in no way impact or affect the substantive content, findings and recommendations in the report,” it stated.

This episode is a stark illustration not only of the potential financial losses that could be incurred by misusing AI, but also the reputational costs. Once you admit to outsourcing your thinking to a machine – and you get it spectacularly wrong – it can take years to earn back trust from your most important stakeholders. Reputationally speaking, AI is already down a deep hole. According to Edelman’s Trust Barometer 2025, less than one in three people in the US are comfortable with businesses using it.

While we tout AI’s biggest advantages as speed and efficiency, in the context of this data, these benefits come at a steep price, especially when we already know that the technology tends to misrepresent information and people. In the realm of thought leadership, meanwhile, imperfection may be acceptable, but inaccuracy never is. When you get it wrong, the cost to you and your business or career is high. You risk losing trust, which is what your success ultimately rides on.

Human-written over AI-driven thought leadership

True thought leadership brings together expertise and experience. It is rooted in context and the complexity of human judgment. While AI may be excellent at replicating knowledge, it cannot convincingly fake nuance and understanding.

Neither can it impersonate us convincingly enough. Voice is not just how we sound, it’s also how we think. It shows the world our culture and identity – from our lexicon and sentence structures to the points we choose to emphasise, include or exclude.

Our individual perspectives and ideas are shaped by everything that makes us human – our upbringing, communities, biases, and beliefs. These influences colour how we interpret information and communicate ideas. They make us relatable, or funny, or quirky, or deep, or interesting. And they make us real.

This convergence of credibility, relatability, and originality creates a powerful foundation for thought leadership, one that blends authority, likability, and trust to build a truly valuable and influential brand. AI fails this framework because it draws from a finite database of existing material. It adds nothing new. Human thought leadership, by contrast, is defined by storytelling and the ability to connect dots in novel ways, creating meaning and offering value.

In a world where everyone has access to the same information, our only real advantage is who we are. Personality is our last remaining differentiator.

The limits and value of AI

However, rejecting AI for thought leadership doesn’t mean rejecting it entirely. The smartest leaders are learning to use AI as a tool instead of a replacement. It can assist with research, structure, or editing, but the critical thinking and fact-checking must remain human.

We’re also finding, through various studies, that constantly turning to AI leads to cognitive atrophy. In other words, the more we delegate our thinking to machines, the more we become unable to do it for ourselves.

Sabine Kühn is the co-founder and CEO of Talent & Truth

Read more: Data quality is no longer optional