Ray Eitel-Porter led Accenture’s global responsible AI practice and co-authored Governing the Machine. This episode asks what happens when AI moves from giving answers to taking actions — and why AI Governance may yet become the cornerstone of management in the Age of Intelligence.

Chatbots make mistakes; agents do things. Today’s AI can hallucinate, but Agentic AI spends money, updates systems, contacts third parties and triggers workflows. When it fails, it will fail at speed with potentially huge consequences.

A company-destroying AI accident is plausible. Ray does not dismiss the risk of a mid-sized business being badly damaged by an AI agent running loose — deleting records, spending cash, exposing data or creating operational chaos before anyone notices.

“Human in the loop” is not a magic shield. Humans only help if they understand the system, stay alert and have authority to intervene. The more accurate AI becomes, the easier it is to over-trust it.

If AI is right 99% of the time who stays awake for the 1%? The smarter the AI, the more dangerous complacency becomes. Ray highlights the example of “cognitive speed bumps” — deliberate pauses that force people to question the machine rather than simply approve its output.

Big firms are waking up; smaller firms may be exposed. Regulation, reputational risk and financial losses are pushing large corporates towards better AI governance. Mid-sized companies may adopt powerful tools faster than they build controls.

AI regulation will probably be sector by sector. Healthcare, finance, recruitment and public services need different rules because AI risk depends on context. One global rulebook is unlikely.

The best governance uses existing business systems. Do not bolt on a new AI bureaucracy. Review procurement, compliance, risk, legal and operational controls with an AI lens.

Accountability should sit with the business owner. Not just IT, legal or the vendor. The executive who wants the AI system and signs off the business case should own both the upside and the risk.

Boards need to ask three questions. Who is accountable? Do we know where AI is being used? Have people been trained well enough to use it safely and productively?

Training is where many rollouts fail. Generic Copilot training is not enough. The value comes when teams redesign real work around AI, not when they learn a few prompts.

Governance may become a source of value. In a world where machines produce more of the work, customers will pay for assurance: that outputs have been checked, systems are controlled and someone trustworthy stands behind them.

The more we worry about the technology of AI, the more we come back to people. Ray’s message is cautiously optimistic: AI governance is possible. But the weak link may yet be the humans.

See omnystudio.com/listener for privacy information.

Olivier Toubia is a Columbia Business School professor working at the frontier of AI, marketing, consumer behaviour, and digital twins. This was recorded late last year but his arguments are relevant and clear: LLMs drive towards the average which means its hard to mimic the diversity of human existence.

• Digital twins are promising — but not ready. Synthetic consumers for research purposes (e.g. virtual focus groups) are fast and cheap, but Toubia’s research suggests they are still too biased and inconsistent to trust too far. Learn what happened when they compared digital and human twins' responses to the same questions.

• Politics is the worst domain to deploy synthetic focus groups: The base model exerts a powerful pull. Twins often reflect the worldview of the model as much as the individual they are meant to represent, especially in politics, where performance is weakest. AI is too sensible - compared to humans it tends to be more moderate: in one study 45% of humans wanted to deport illegal immigrants and only 5% of digital twins agreed.

• Synthetic personalities reflect a clear pro-tech bias. Twins tend to be more comfortable with algorithms, less concerned about privacy, and more trusting of technology than humans actually are. AI twins thinks AI is smarter than humans....

• Generative AI has entered its monetisation phase with advertising as the likely endgame Subscriptions and enterprise sales matter, but the bigger prize is becoming the interface for search, discovery, decision-making, and purchase.

• Business adoption remains uneven but labour anxiety is rational Many pilots are failing to deliver meaningful returns; senior leaders are often enthusiastic, while middle managers and employees are more sceptical.

• AI can be creative, but not at the edges. AI tools can recombine ideas - "its a fallacy to think that AI will never be able create new things". However, it still misses the diversity of lived experience that gives humans their edge. The push is towards the average.

We usually worry about AI driving polarisation - Olivier's work shows that current LLM tools trend towards the opposite. Humans can provide the spikiness and AI the padding.

See omnystudio.com/listener for privacy information.

Cristina Caffarra is an eminent economist and veteran antitrust practitioner who has become a leading voice behind Eurostack — a grassroots, industry-led push to rebuild Europe’s digital infrastructure as a sovereignty and competitiveness play. She has a strong argument to make:

• Europe is a “digital colony” — and it’s self-inflicted. US hyperscalers are excellent; Europe vacated the field through fragmentation, weak risk capital, and complacency.

• The "kill switch" is a distraction; dependency is the disease. The real risk is gradual denial: deprecated features, constrained access, and strategic leverage — not a Hollywood blockbuster blackout.

• Productivity is the core argument. Europe’s gap is investment per worker, especially into high-growth tech that diffuses across the whole economy.

• Regulation can’t create an industry. Antitrust and platform rules “nibble at the corners”, take years, and leave the giants stronger — while absorbing all the political oxygen.

• Europe chose theatre over building. “Taming Big Tech” became a substitute for the only question that mattered: where are Europe’s builders, customers, and scale-ups?

• Demand is the lever, not more grants. Without customers, no stack survives — procurement and enterprise buying decisions are the flywheel.

• Procurement should be the no-brainer. Every other major power has local preference norms; Europe’s non-discrimination logic is now being weaponised against European options. She notes that even the European Commission's own CIOs focuses on performance and efficiency alone.

• Private enterprise is the real swing voter. Public sector is ~20% of demand; the other 80% sits with CEOs and CIOs who complain about European weakness — and then buy American.

• European tech can compete cost — but not ease of use. European components exist; what’s missing is end-to-end “peace of mind” and the glue between parts.

• Mercedes is the case study. They want sensitive AI loads (e.g. autonomous driving) not wholly dependent on US infrastructure — but need suppliers and buyers to co-design workable, integrated alternatives.

• She argues that no one wants to “decouple” form the US — that’s a straw man. The practical goal is share: move European supply from <20% to something like 30–40% in a growing market.

• This is wartime logic, not business-as-usual. Europe has surged before under pressure; Caffarra argues that its time to stop waiting for Brussels and to start acting like a superpower.

She leaves us with a blunt challenge: will Europe keep buying convenience — or invest, through demand, in a tech stack that keeps its innovation future in its own hands?

See omnystudio.com/listener for privacy information.