The AI Mistake That's Costing Companies Billions

Jan de Vries · 2026-09-16

AI is hitting boardrooms so fast, it's leaving many executives scrambling. But according to Jim Harris, a leading GenAI and disruptive innovation specialist, the real danger isn't moving too slow. It's slapping new technology onto old ways of working and expecting magic to happen. Harris has advised executive teams at giants like American Express, Barclays, IBM, SAP, and Walmart. He was even named Speaker of the Year by TEC Canada for his AI work. In this conversation, he breaks down why a complete redesign is the only path to real transformation. ### How to Introduce AI Without Spooking Your Team First things first, you've got to create a sense of psychological safety. Harris points out that most people in developed nations are genuinely afraid of AI, mainly due to job loss fears. As adults, we're just wired to be uncomfortable with new ways of working. So, what's the playbook for leaders? - **Guarantee job security.** Make it clear that AI is here to augment, not replace. - **Provide comprehensive training.** Don't just drop new tools and run. - **Frame it as a skills upgrade.** Show how AI will enhance their employability and value. It sounds simple, but organizations have to work hard to communicate this. It's about building trust, not just deploying software. ### Why "Sprinkling AI" on Old Processes Fails Here's a story that says it all. Harris was at CES in Las Vegas and spoke with the head of AI for Ernst & Young in the US. EY, with its 400,000 employees, has invested a staggering $1 billion in AI. They had a classic 46-hour risk management process for clients. Their first thought? "Let's apply some AI to it." Working with the original process designers, they got it down to 32 hours—a solid 20% improvement. Then they tried something radical. They brought in fresh, tech-first employees and paired them with risk managers who'd never touched the old process. The result? They slashed it to **15 minutes**. Of course, being EY, they had to add back compliance and verification steps. The final time? Six hours. That's still an over 80% reduction from the original 46. "You can't just sprinkle some AI fairy dust on your existing processes," Harris warns. "You really need to start with a blank sheet of paper." The people who built the old system are often too close to it. You need new eyes to ask, "If we designed this today from scratch, what would we do?" ### Beyond ChatGPT: AI's Transformational Potential The gains go far beyond chatbots. Harris highlights the work of pioneers like Demis Hassabis of Google DeepMind. The potential is in reimagining entire workflows, not just automating single tasks. The key takeaway? True AI transformation requires courage. It means being willing to scrap what you know and redesign from first principles. It's messy, it's challenging, but as the EY example shows, the payoff isn't just incremental—it's revolutionary. You free up your people for the strategic, human-centric work that machines can't do. That's where the real value gets created.