Who's in the room when AI gets built? Interview with Luli Adeyemo

Who's in the room when AI gets built? Interview with Luli Adeyemo

In this episode of Mindful AI, we welcome Luli Adeyemo, Executive Director of TechDiversity Foundation, co-founder of Tendertrace, and founder of Best Case Scenario. Her three decades in technology have made her one of Australia's most distinctive voices on AI governance and inclusion. From winning Britain's 1986 BMX World Championship as a teenage Black girl in a white-dominant sport to building Australia's first Diverse AI Governance Practitioners Programme, Luli brings lived experience and commercial rigour to a conversation the industry is only just beginning to have properly. We explore the "influence gap": what happens when the people most affected by AI aren't in the room when it's designed, and why governance is the boring-but-necessary foundation we can't afford to skip. The conversation moves through real failures (a University of Sydney study on melanoma-detection AI under-performing on darker skin), the international standards most organisations don't know exist (ISO/IEC 42005), and the widening gap AI risks creating between the haves and have-nots. Luli argues that diversity isn't something you bolt on at the end. It has to be the methodology from the start. If you care about who gets to shape AI and who pays when it fails, this conversation is for you. Links from the episode: * Luli Adeyemo [https://www.linkedin.com/in/luliadeyemo/] on LinkedIn * TechDiversity Foundation [https://tech-diversity.com.au/] * Australia's First Diverse AI Governance Practitioners Programme [https://tech-diversity.com.au/australias-first-diverse-ai-governance-practitioners-programme/]: * Dr Kobi Leins [https://www.kobileins.com/] * Cultural Infusion [https://culturalinfusion.com/about-us/] * Diversity Atlas [https://atlas.culturalinfusion.com/] * University of Sydney skin cancer AI study [https://www.theguardian.com/society/2021/nov/09/ai-skin-cancer-diagnoses-risk-being-less-accurate-for-dark-skin-study] (via The Guardian) * The Smith Family on breaking the poverty cycle through education [https://www.theaustralian.com.au/sponsored/fPaP4hMf5xXFgUnzmVPX/breaking-the-poverty-cycle-through-education/⁠] CHAPTERS: (00:00) Welcome and intro to Luli (02:00) BMX World Champion, 1986 (04:00) Why diversity in tech matters (07:00) Diversity Atlas and TechReflects (10:00) When AI gets melanoma wrong (13:00) Diversity as methodology (15:00) The Influence Gap (18:00) Why governance is necessary (20:00) The AI Governance Programme (22:00) Velocity vs governance (27:00) The growing AI divide (30:00) TechDiversity Awards 2026

AI’s Reality Problem: Why Maps May Be the Key to Trustworthy AI | Kirk Mitchell

In this episode of Mindful AI, we sit down with Kirk Mitchell, a Google geospatial leader and long-time champion of diversity, equity, and inclusion, to explore what happens when AI is grounded in the physical world.Kirk traces his journey from hand-drawn cartography at RMIT to today’s planet-scale mapping systems, and explains why geospatial context is becoming a core ingredient for more reliable AI. We unpack “grounding” (connecting generative models to trusted, up-to-date maps data), and why this matters for everything from everyday navigation to reducing hallucinations in location-based queries.The conversation moves from maps as convenience to maps as human rights infrastructure, including how tools like Plus Codes can function as an address for communities that do not have formal street addressing. We also discuss what “local” really means in the Global South: two-wheel mobility, landmark-based directions, and safety-first design for riders.From there, Kirk shares some of the most compelling “AI for good” work happening right now:Project Green Light, using AI-driven recommendations to improve traffic signal timing and reduce emissions (including work in Boston).Contrail avoidance, where AI-informed routing can significantly reduce warming impacts from aviation contrails.Earth Engine, which makes large-scale Earth observation and analysis accessible for research and many noncommercial uses.We also go into the more nuanced terrain: autonomous vehicles and safety (and the second-order impacts on professional drivers), bias and perspective-taking, and Kirk’s caution that outsourcing creativity and spatial reasoning may come with hidden cognitive costs.If you’re interested in AI that is not just powerful, but situated, accountable, and human-centered, this episode offers both practical frameworks and philosophical pressure-tests.🔔 Subscribe to Mindful AI for more in-depth conversations on ethical AI and human flourishing.💬 Do you think AI will expand human creativity, or slowly replace the struggle that makes creativity meaningful? Tell us in the comments.🎙 Links from the episode:🔗 Grounding with Google Maps [https://blog.google/innovation-and-ai/technology/developers-tools/grounding-google-maps-gemini-api/] (Gemini / Maps grounding):🔗 Project Green Light [https://sites.research.google/gr/greenlight/] (traffic signal optimization)🔗 Project Contrails [https://sites.research.google/gr/contrails/] (contrail avoidance research)🔗 Plus Codes [https://maps.google.com/pluscodes/] (addresses without street addresses)🔗 Google Earth Engine [https://earthengine.google.com/] 🔗 Waymo safety benchmarks [https://waymo.com/blog/2023/12/waymo-significantly-outperforms-comparable-human-benchmarks-over-7-million] (7M+ rider-only miles) 🔗 Book: The Second Machine Age [https://wwnorton.com/books/the-second-machine-age/] (Brynjolfsson & McAfee) 🔗 Book: Could Should Might Don’t [https://www.allenandunwin.com/browse/book/Nick-Foster-Could-Should-Might-Don%27t-9781837263837] (Nick Foster): Chapters: 00:00 Welcome, Tech Diversity Awards, and Kirk’s DEI roots 03:00 From cartography to Google: mapping at planet scale 05:34 Why AI is essential for high-definition, always-updated maps 06:17 Grounding AI in the physical world (and reducing hallucinations) 08:45 Addresses as a human right: Plus Codes and inclusion 10:05 Local navigation realities: two-wheelers, landmarks, and safety 13:11 Democratizing geospatial AI: prompts, “vibe coding,” and new builders 15:15 Earth Engine: geospatial AI for climate and sustainability 18:03 Maker tools vs autonomy: Waymo and the safety promise 23:23 Project Green Light: smarter traffic signals, lower emissions 25:12 Aviation contrails: AI-informed routing for climate impact 26:57 Mapping kelp forests: climate resilience in Australia 33:00 Bias, tribalism, and AI as a thinking partner 48:11 Practical advice: mindful AI use, diverse voices, and “shipwrecks”👉 Enjoyed this episode? Like, comment, and subscribe to support mindful, meaningful conversations on AI and ethics.

AI's Blind Spot: Why We Need Ubuntu in Technology | Nazareen Ebrahim

In this episode of Mindful AI, we welcome Nazareen Ebrahim, South African AI ethicist and founder of Naz Consulting International, whose work champions diverse voices in the global AI conversation. From a pivotal conversation at Web Summit 2018 to mentorship with the World Economic Forum, Nazareen now educates healthcare professionals, schools, and communities across Africa about responsible AI. We explore why diverse representation matters, how Western-centric training data perpetuates bias, and what it means when one in three South Africans still don't know about AI. The conversation tackles solutions in action: South Africa's AI maturity framework, healthcare regulations for practitioners, and Nazareen's forthcoming AI for Everyone Handbook. She argues for agency through education, better questions, and the African philosophy of Ubuntu—"I am because you are." If you believe AI should reflect all of humanity, not just Silicon Valley, this conversation offers both critique and practical pathways forward. 🎙 Links from the episode:🔗 Centre for AI and Digital Policy: https://www.caidp.org/ [https://www.caidp.org/]🔗 Lelapa AI (South African language models): https://lelapa.ai/ [https://lelapa.ai/]🔗 Centre for Humane Technology: https://www.humanetech.com/ [https://www.humanetech.com/] CHAPTERS: 00:00 Skateboarding Johannesburg and the journey to AI ethics 05:00 Web Summit 2018: the conversation that changed everything 10:00 Why diverse voices matter in AI ethics 18:00 Western-centric training data and the mirror problem 24:00 Indigenous AI research and language equity 34:00 Centre for AI and Digital Policy: building AI literacy 41:00 Solutions in action: education, maturity frameworks, and healthcare regulations 50:00 What we can do: asking questions and responsible digital citizenship 53:00 Ubuntu: AI can only be as mindful as the humans behind it

# 6 AI Ethics Is Being Ignored, Can We Fix It In Time? Interview with Mike Spaeth

In this episode of Mindful AI, we speak with Mike Spaeth, Global Vice President at the United States Artificial Intelligence Institute (USAII). Mike’s journey spans politics in Washington DC, early AI personalization at Amazon, IBM’s Watson, and now leading global AI education across 160+ countries. We explore the AI race, ethics, and education: from empowering young students in Nigeria and Ghana, to vendor-agnostic certifications, to the question of whether acceleration is overshadowing accountability. Mike shares why inclusivity, ethics, and AI literacy are essential if we want a future where technology serves human flourishing. If you’re interested in AI for good, this conversation is packed with insight and inspiration. 🔗 Links from the episode: • United States Artificial Intelligence Institute (USAII) Chapters 00:00 – Welcome & Mike’s birdwatching practice 02:30 – What the USAII is and why ethics matters 07:00 – The global AI race and education in Africa 10:00 – Mike’s journey from politics to Amazon & Watson 13:00 – Jobs, dignity, and what humans do best 17:30 – Making AI literacy inclusive & jargon-free 21:30 – Ethics training for students, engineers, and C-suites 26:00 – Acceleration, military pressure, and healthcare promises 33:00 – What listeners can do today to influence AI’s future 38:00 – The literacy gap in HR and schools 44:00 – Energy, infrastructure, and affordability 49:00 – Why Big Tech holds the keys 52:00 – Black boxes, explainability, and risks 59:00 – Positive use cases of AI and closing reflections Follow Mindful AI for more in-depth conversations on ethical AI and human flourishing.

#5 Will AI solve cancer? Interview with Mary Adewunmi

In this episode of Mindful AI, we speak with Mary Adewunmi, a rising force in medical AI whose work is transforming cancer diagnostics and chronic disease treatment. Mary is a PhD candidate at the Menzies School of Health Research, founder of CARES AI, and a recipient of multiple prestigious awards for her deep learning research. But behind her technical achievements is a deeply personal story: a tragic misdiagnosis in her own family sparked her journey into healthcare AI. We explore the promise—and the pitfalls—of AI in medicine: from democratizing diagnosis to the urgent need for ethical oversight, bias mitigation, and human-centered design. Mary shares how empathy, co-design with communities, and data accountability must guide the future of AI in healthcare. If you’re passionate about AI for good, this conversation will leave you inspired—and better informed. 🔔 Subscribe to Mindful AI for more in-depth conversations on ethical AI and human flourishing. 💬 What role should AI play in healthcare? Let us know in the comments! 🎙 Links from the episode: 🔗 CARES AI: https://caresai.org/ [https://caresai.org/] 🔗 Menzies School of Health Research: https://www.menzies.edu.au/ [https://www.menzies.edu.au/] 📌 Chapters 00:00:00 Introducing Mary Adewunmi 00:00:45 A personal tragedy sparks a mission 00:03:00 From computer science to cancer care 00:07:00 Why cancer research became her focus 00:10:00 What CARES AI is doing differently 00:12:00 AI outperforming doctors: promise and pitfalls 00:15:00 Empathy and accuracy in AI models 00:17:00 Democratizing healthcare with AI 00:20:00 The role of AI in treatment and clinical support 00:24:00 Ethical concerns: bias, black boxes, data privacy 00:28:00 Policy gaps and real-world consequences 00:33:00 Human-in-the-loop and co-design with patients 00:38:00 Can AI be truly creative? 00:45:00 Final thoughts and what needs to change 00:46:30 Subscribe & stay tuned for Part 2 👉 Enjoyed this episode? Like, comment, and subscribe to support mindful, meaningful conversations on AI and ethics.