80% of data and AI products in organisations are never used. That is not a pilot problem; it is an AI value management crisis happening at scale right now.In Episode 12 of The AI Values Podcast, Edosa Odaro and Lindley Gooden sit down with Nicolas Averseng, Chief Product Officer at DataGalaxy and founder of Yui, one of the world's first dedicated AI value management platforms. This is one of the most practically uncomfortable conversations the podcast has produced. Nikola introduces what he calls "the fallacy of efficiency": the idea that the 30% productivity gains promised by agentic AI and GenAI deployments rarely materialise once they meet a CFO's scrutiny, and he explains why this trap catches even the most capable organisations. If your AI business case rests on efficiency gains, AI risk management, or responsible AI governance, this episode is required listening for your board.ABOUT NICOLAS AVERSENG:Nicolas Averseng is Chief Product Officer at DataGalaxy. He previously founded Yui, one of the world's first value management platforms for data and AI investments, which was acquired by DataGalaxy. He specialises in transforming data and AI governance from a compliance exercise into a genuine catalyst for value creation, and works with senior leaders globally to close the gap between AI investment and AI ROI.◼ Nicolas Averseng: https://www.linkedin.com/in/naverseng/WHAT IS COVERED:► The 80% problem: why the majority of data and AI products in organisations are never adopted, and the root cause most CDOs and boards refuse to acknowledge until it is too late► The fallacy of efficiency: how the promise of AI-driven productivity gains disappears when it meets the CFO's budget review, and why efficiency alone is the wrong objective for responsible AI implementation► The foundation-building trap: why organisations spend years building data platforms nobody uses, and how outcome-led AI implementation and minimum viable use cases change the equation entirely► AI value versus AI risk: Nikola's core argument, captured in his own words ("There is no value story without the downside"), explains why AI governance and AI risk management are not separate conversations, and how the EU AI Act is forcing this realisation even on organisations that resist it► The 14-month CDO problem: why chief data officers burn out before they deliver, what the average tenure of 12–24 months tells us about the state of AI value management, and what genuinely fixes it⏰ EPISODE TIMESTAMPS:00:00 — Cold open: years building a data platform nobody could use00:45 — Lindley introduces the 80% stat, what is it really costing us?04:49 — The foundation-building trap: why organisations build before they think08:15 — Starting from the end: the minimum viable AI use case approach12:09 — What does "value" actually mean? ROI vs. genuine user benefit16:40 — AI's wider cost: employment, ESG, client relationships, trust20:19 — The fallacy of efficiency: why GenAI's 30% promise rarely shows up24:26 — Are organisations getting better? Closing reflections──────────────────────────────────────ABOUT THE AI VALUES PODCAST:The AI Values Podcast is where leaders come to think clearly about the trade-offs behind AI adoption not just the opportunities. Hosted by Edosa Odaro (author, 'The Values of AI') and Lindley Gooden (author, 'The Future of Truth'), with weekly conversations at the intersection of AI, trust, governance, and the future of work.🎙 SUBSCRIBE to The AI Values Podcast for honest, rigorous conversations at the intersection of AI ethics, AI governance, and business leadership.◼ Find out more: https://www.theaivalues.org ◼ Reach out: podcast@theaivalues.org ◼ Get the Weekly AI Values Dispatch → https://pages.theaivalues.org◼ Edosa Odaro: https://www.linkedin.com/in/edosa/◼ Lindley Gooden: https://www.linkedin.com/in/lindleygooden/

What if your organisation's AI strategy rests on a system that doesn't understand the world the way you think it does?

Edosa Odaro and Lindley Gooden go head-to-head on Episode 11 of The AI Values Podcast on the debate that is quietly splitting the AI research world in two: LLMs versus world models, pattern recognition versus causal understanding, and the governance question that follows from the answer. This is not a technical forecasting conversation. It is the one your board should have before the next autonomous AI budget line gets signed.

WHAT IS COVERED:

► "The biggest problem around AI is that we speak about it as if it actually understands the world": why conflating language fluency with genuine understanding is the most dangerous assumption in AI governance, and what world models offer instead

► The transport company case study: an LLM correctly identifies a flood and reroutes an entire fleet, then causes gridlock, because it cannot model the downstream consequence of its own decision

► LLMs are not enough": why Lindley argues that without physical awareness, AI cannot be trusted in medical tech, sports science, or any domain where the real world pushes back

► Do AI systems need arms and legs? Why giving autonomous AI physical form is not an engineering decision but a values question, and why Silicon Valley's race to remove guardrails is precisely the moment human oversight matters most

► "I feel that we're giving human expectations to non-human systems": the framing that redefines AI alignment as a relationship risk, and what that means for every organisation deploying AI at scale

⏰ EPISODE TIMESTAMPS:

00:00 — The AI illusion: mistaking language fluency for genuine understanding

01:06 — LLMs vs world models: what the difference actually means in practice

02:09 — AGI and ASI: which architecture leads to general intelligence?

03:28 — Arms, legs, and autonomy: the embodied AI question nobody wants to answer

04:54 — Case study: the flood, the fleet, and the limits of LLM reasoning

06:35 — Silicon Valley, fewer guardrails, and the autonomy debate

08:03 — Causality, consequence, and what AI needs to understand an ecosystem

09:17 — "We're giving human expectations to non-human systems"

10:20 — Augmentation or transformation? The values-first answer

12:12 — A direct challenge: have this conversation in your own organisation

ABOUT THE AI VALUES PODCAST:

The AI Values Podcast is where leaders come to think clearly about the trade-offs behind AI adoption not just the opportunities. Hosted by Edosa Odaro (author, 'The Values of AI') and Lindley Gooden (author, 'The Future of Truth'), with weekly conversations at the intersection of AI, trust, governance, and the future of work.

🎙 SUBSCRIBE to The AI Values Podcast for honest, rigorous conversations at the intersection of AI ethics, AI governance, and business leadership.

◼ Find out more: https://www.theaivalues.org

◼ Reach out: podcast@theaivalues.org

◼ Get the Weekly AI Values Dispatch → https://pages.theaivalues.org

◼ Hosts: Edosa Odaro & Lindley Gooden

Most AI projects fail not because the technology fails, but because humans never fully integrate it into their way of working.

Amy unpacks what successful AI implementation really looks like, not from the tech side, but from the human side. From boardroom tensions to culture change, career disruption to lifelong learning, this is the conversation most AI podcasts aren't having.

THIS WEEK: A GUEST EPISODE

Our guest is Amy Shi-Nash, Professor of AI Practice at Monash University and Co-founder & CEO of Occupy. Amy brings more than two decades of hands-on implementation experience across global commercial organisations and academia — building AI that actually changes behaviour, not just processes. She is one of a rare group of practitioners who hasseen both sides: what makes AI succeed at scale, and why the human side is the part most leaders get wrong.

◼ Connect with Amy

⏰ EPISODE TIMESTAMPS:

00:00 — Cold open: "AI changes the shape of work, this is bigger than they say"

01:40 — Why 85–95% of AI projects fail to deliver value

05:00 — Meet Amy Shi-Nash: 25 years making AI work in large organisations

06:10 — What good AI implementation actually looks like (from the human side)

08:40 — The junction between humans and AI: why it determines everything

11:00 — From faster to different: culture change as the real measure of success

13:30 — Real-world examples: users building agents, evolving the system themselves

14:25 — Boardroom tensions: innovators, sceptics, literacy gaps, and time horizons

17:30 — Are we ready? The readiness question every board avoids

19:50 — Consequences: intended, unintended, and unevenly distributed

23:00 — The literacy imperative: AI education across every role and level

25:55 — Job composition change: what AI does and what humans must do

27:45 — Distribution of value: how AI reshapes the organisation's shape

29:00 — Flatter structures, specialist roles, fractional work, and the rise of the entrepreneur

32:00 — Debrief: culture, career paths, and the age of the specialised side hustle

ABOUT THE AI VALUES PODCAST:

The AI Values Podcast is where leaders come to think clearly about the trade-offs behind AI adoption not just the opportunities. Hosted by Edosa Odaro (author, 'The Values of AI') and Lindley Gooden (author, 'The Future of Truth'), with weekly conversations at the intersection of AI, trust, governance, and the future of work.

🎙 SUBSCRIBE to The AI Values Podcast for honest, rigorous conversations at the intersection of AI ethics, AI governance, and business leadership.

◼ Find out more: https://www.theaivalues.org

◼ Reach out: podcast@theaivalues.org

◼ Get the Weekly AI Values Dispatch → https://pages.theaivalues.org

◼ Co-Hosts Edosa Odaro & Lindley Gooden