Synopsis: In this Techstrong AI video interview, Unisys President and COO Mike Thomson dives into the challenges organizations are encountering as they look to operationalizing artificial intelligence (AI).
In this Techstrong AI interview, Mike Vizard sits down with Michael Thomson, president and incoming CEO of Unisys, to explore the most disruptive enterprise technologies heading into 2025. Thomson notes that while generative AI (GenAI) is still in a heavy discovery phase, some applications—such as automation in IT operations, marketing, and service delivery—are already delivering real value. He sees two primary vectors of AI adoption: using it to enhance internal processes and embedding it into revenue-generating products. Industries like finance and healthcare are leading the way in commercializing AI, while others are still experimenting. Thomson also highlights that the hype is giving way to practical deployments, with organizations focusing on improving performance, cost-efficiency, and customer experience.
Thomson emphasizes that enterprises are now shifting from a fear-of-missing-out mindset to a strategic approach where AI capabilities are becoming table stakes. The future of AI, he says, will involve integrating it more deeply into workflows—not just as a tool but as a digital teammate. He imagines a future where AI anticipates user needs without being prompted, delivering seamless experiences, like automatically configuring smart meeting rooms. However, he also acknowledges the tension between probabilistic AI models and the deterministic nature of many business processes. To address this, companies are exploring the use of smaller, purpose-built language models (SLMs) deployed at the edge, supported by solid data models and security frameworks to avoid hallucinations and misuse.
Security remains a top concern as AI becomes more prevalent. Thomson warns of rising threats from cybercriminals exploiting AI agents, citing potential scenarios like facial recognition hacks. He also raises alarms about the looming impact of quantum computing, calling it a “Y2K-style” event. Once quantum capabilities reach a certain point, current encryption methods could be rendered obsolete, posing a massive security risk unless organizations act now. Companies must map vulnerabilities, prepare both software and infrastructure defenses, and adopt a “quantum-safe” strategy well before 2028. In closing, Thomson urges organizations to take a practical, phased approach to AI adoption and be proactive about security—because when these technological shifts hit full force, it may be too late to react.
