A recent collaborative event between Ergo and ADAPT at Trinity College Dublin aimed to unite business, academia, and cybersecurity to explore the opportunities and challenges posed by generative AI. Featuring thought leaders from Trinity College Dublin, Microsoft, ServiceNow, and the University of Pittsburgh, the event offered clarity on the deployment, risks, and promise of next-generation AI systems.
Opening the day, Professor John Kelleher, Chair of Computer Science and Director of ADAPT, unpacked the evolution of AI, from symbolic logic and early rule-based systems to today’s generative AI models. He emphasised the need to demystify how models like GPT operate, particularly around language prediction, training methods, and the risks of hallucinations and bias. Kelleher drew attention to the crucial distinction between language mimicry and real-world understanding, underlining the importance of context and prompt engineering.
Ergo CIO Lorne Chedzey highlighted the spectrum of generative AI deployment models – open source vs. proprietary, on-premise vs. cloud – and broke down their respective trade-offs. He stressed that successful integration depends not on flashy tech, but on strong use-case alignment and enterprise readiness. Chedzey also explored “agentic AI” and how tools like Microsoft Copilot or ServiceNow’s AI agents vary in complexity from simple retrieval functions to autonomous task execution.
A panel discussion featuring Dale Cheeseman (ServiceNow), Jeevantika Lingalwar (Microsoft), and Steve Blanche (Ergo) discussed real-world implementation. Topics ranged from change management to managing hallucinations and fine-tuning models for internal business tasks. The panel underscored that while technical capabilities are expanding rapidly, trust and governance remain key to driving enterprise adoption.
Finally, bringing a transatlantic perspective, Professor David J. Hickton of Pitt Cyber at the University of Pittsburgh addressed the growing imperative to defend AI systems against misuse and cyber threats. “We are building the digital backbone of society, its integrity is non-negotiable,” Hickton warned. He presented strategies to secure AI systems, including policy frameworks and human-in-the-loop accountability structures.
The event closed with a clear message: AI maturity is still nascent across most sectors. According to McKinsey and Stanford research cited during the sessions, only 1% of organisations can claim full AI integration. But with thoughtful design, clear governance, and responsible innovation, that number is poised to grow.
