Ayush Prakash Podcast

AI could expand access to mental-health support, but a convincing conversation doesn’t establish that a chatbot is safe. Tristan Glatard explains why mental-health AI needs evidence, evaluation and human supervision, and how its potential extends from conversational tools to predicting treatment responses and identifying crises.

Tristan is Scientific Director of the Krembil Centre for Neuroinformatics at the Centre for Addiction and Mental Health (CAMH) in Toronto. His background is in computer science. In this conversation, he examines where AI could help patients, where general-purpose chatbots need limits, and why privacy, patient perspectives and clinical oversight matter.

In this conversation:

What mental health means across the lifespan

How AI could help match patients with treatments

Why mental-health AI includes wearables and predictive tools

The difference between healthcare chatbots, wellbeing apps and general-purpose AI

The case for support when a human therapist isn’t available

Whether people disclose more to chatbots, and what remains uncertain

Why wellbeing tools need evaluation

How hospital researchers balance opportunity, safety and health equity

Why crisis detection and data privacy matter

How patients should shape the use of AI in their care

Who evaluates and supervises AI support conversations

A surprising finding from Tristan’s ongoing image-analysis research

What discovery, collaboration and good work mean to him

Chapters:

00:00 AI and Mental Health at CAMH

01:29 What Do We Mean by Mental Health?

03:41 How Technology Can Help Personalize Treatment

05:48 AI Beyond Chatbots: Wearables and Prediction

08:58 The Case for an AI Companion

13:15 Do People Open Up More to Chatbots?

15:03 Where AI Mental-Health Support Breaks Down

16:56 Balancing Opportunity, Safety and Health Equity

20:00 Crisis Detection and Data Privacy

22:12 The AI Healthcare Conversation We Need

26:28 Who Supervises AI Support?

30:19 A Surprising Research Discovery

33:03 What Keeps a Researcher Going

35:40 What Gives Tristan Hope

37:30 Where to Follow Tristan’s Work

Connect with Tristan:

Website: https://www.camh.ca/en/science-and-research/science-and-research-staff-directory/tristanglatard

LinkedIn: https://www.linkedin.com/in/tristanglatard/

Podcast:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG  

Apple Podcasts: https://apple.co/3qXL37W  

Connect:

Website: https://ayushprakash.com  

LinkedIn: https://www.linkedin.com/in/prakash-ayush/  

Instagram: https://instagram.com/ayushprakashofficial  

Book:

https://www.amazon.com/dp/0981182135

A Canadian frontier model alone won’t give Canada control over its AI future. Anna Jahn explains why AI sovereignty also depends on data, energy, compute, talent, and governance, and why countries may need to combine their strengths to reduce dependence on the biggest AI powers.

Anna is Executive Director of the Centre for Media, Technology and Democracy at McGill University. She previously led public policy and inclusion work at Mila. In this conversation, she examines who holds power over AI now, why Canadian research strength hasn’t produced the same economic strength, and what a credible Canadian approach could look like.

In this conversation:

Why researchers, the public, government, and industry talk past one another

Whether democratic policymaking needs to match the pace of AI

Why voluntary company safeguards aren’t enough

The tension between promoting AI adoption and protecting the public

Why a domestic frontier model is only one part of AI sovereignty

How Canada could work with other middle powers

Why Canadian AI talent often builds companies elsewhere

What Anna wants to hear from Mark Carney about AI

Montreal’s opportunity to offer a different vision for AI

What it would take for Canadians to trust and benefit from the technology

Chapters:

00:00 Who Gets a Say in AI Policy?

02:19 Can Governments Keep Up With AI?

05:17 Who Is Responsible for AI Safety?

08:03 AI Adoption vs. Public Protection

11:15 Why One Model Won’t Deliver AI Sovereignty

14:07 Why Canada Struggles to Commercialize AI Research

16:49 What a Middle-Power AI Alliance Could Look Like

21:55 What Canada Needs to Do First

24:44 Is Montreal Still a Leading AI Hub?

27:38 Should Montreal Try to Become Another Silicon Valley?

29:34 What AI Success Would Look Like for Canadians

32:28 What Gives Anna Hope

33:32 Where to Find Anna’s Work

Connect with Anna:

Website: https://mediatechdemocracy.com/en/team/anna-jahn/

LinkedIn: https://www.linkedin.com/in/anna-jahn-8a318241/

Podcast:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG  

Apple Podcasts: https://apple.co/3qXL37W  

Connect:

Website: https://ayushprakash.com  

LinkedIn: https://www.linkedin.com/in/prakash-ayush/  

Instagram: https://instagram.com/ayushprakashofficial  

Book:

https://www.amazon.com/dp/0981182135

Nicholas Nadeau argues that the AI most people use is a bootleg of human expertise: a confident statistical average built from scraped public data, without the judgment, consent, or unpublished knowledge that makes an expert worth consulting.

Nicholas is the CTO and Co-Founder of Onix. He holds a PhD in human-robot interaction and previously served as CTO of humanoid-robotics company 1X and data company SmartOne.ai. Onix is building private AI systems grounded in the approved work and judgment of individual experts.

In this conversation:

■ Whose answer you receive when you ask AI a question

■ Why frontier models blend textbooks, social media, and internet noise

■ What gets lost when AI attempts to imitate a human expert

■ Why an expert's most important knowledge may never have been published

■ How privacy and on-device AI can improve personal usefulness

■ Whether expert AI will replace professionals or increase their capacity

■ What young founders should understand about risk, stamina, and relationships

■ Why smart investors must contribute more than capital

■ How personal AI could eventually connect with robotics

■ The question Nicholas asks himself every day: should Onix exist?

CHAPTERS

00:00 Whose Answer Is AI Giving You?

04:02 Transactional AI vs. Wicked Problems

06:03 What AI Misses About Expert Judgment

11:51 Dark Data and the Knowledge Experts Never Publish

17:28 How Onix Began

22:42 Advice for Young Founders

28:45 Smart Money and Choosing Investors

32:45 Onix Beyond Health: Robots and Human Expertise

36:59 Will Expert AI Replace Humans?

41:20 Would Nicholas Let His Children Use AI?

46:45 The Question That Keeps Him Up at Night

48:49 What Gives Him Hope

50:45 Where to Find Nicholas and Onix

CONNECT WITH NICHOLAS

Website: https://nicholasnadeau.com

Onix: https://onix.life

LISTEN TO THE PODCAST

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

CONNECT WITH AYUSH

Website: https://ayushprakash.com

LinkedIn: https://www.linkedin.com/in/prakash-ayush/

Instagram: https://instagram.com/ayushprakashofficial

AI FOR GEN Z

https://www.amazon.com/dp/0981182135

Renée Sieber explains why schools and public institutions are adopting generative AI before confronting its effects on learning, judgment, privacy, and civic life. Her central warning: when AI substitutes for the struggle involved in thinking and writing, students may become dependent on systems they were never given a meaningful choice to refuse.Renée is an associate professor at McGill University whose work spans AI, natural language processing, public policy, and civic engagement. We discuss what AI literacy should include, why transparency means little when people cannot act on it, how communities can contest unwanted systems, the history of Luddism, and what a genuinely humanizing technology could look like.In this episode:- Why Renée has gained respect for controlled-corpus deep learning- How social justice separates signal from noise in AI policy- Why individual “choice” fails as a frame for structural AI harms- What governments and communities can do when AI becomes embedded- How generative AI can remove productive struggle from education- Why Renée embraces the original meaning of Luddism- Where local, bounded AI systems may serve real human needs- Why Gen Z’s resistance gives her hopeChapters:00:00 What Renée changed her mind about00:46 Where she gets information about AI01:50 Social justice as the signal in AI policy06:31 Why these harms are difficult to discuss09:33 What AI literacy should actually mean14:54 Can communities reject embedded AI?20:28 Rebuilding civic engagement in Canada23:10 What generative AI is doing to students34:38 Why Renée embraces Luddism35:29 What humanizing technology could look like37:26 Thinking, writing, and the value of struggle39:50 The technology Renée would eliminate41:15 What gives her hope43:30 Where to follow Renée’s workConnect with Renée:Website: https://aifortherestofus.ca/McGill profile: https://www.mcgill.ca/geography/sieberPodcast:Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhGApple Podcasts: https://apple.co/3qXL37WConnect with Ayush:Website: https://ayushprakash.comLinkedIn: https://www.linkedin.com/in/prakash-ayush/Instagram: https://instagram.com/ayushprakashofficialBook:AI for Gen Z: https://www.amazon.com/dp/0981182135

Nidhi Hegde explains why giving the public access to chatbots does not make AI democratic when users have no control over how the systems are built, trained, or deployed. Her central argument is that agency, representation, privacy, fairness, and technical robustness must be treated as connected design and governance problems.

Nidhi is an associate professor of computing science at the University of Alberta, a Fellow at the Alberta Machine Intelligence Institute, and a Canada CIFAR AI Chair. We discuss what trustworthy AI means technically, the unavoidable trade-offs between privacy, fairness, accuracy, and cost, why broad AI regulation targets the wrong unit, and what public participation in AI decisions could actually require.

In this episode:

- How model robustness may produce fairer outcomes

- What privacy, fairness, robustness, and trust mean in practice

- Who currently decides the trade-offs built into AI systems

- Why AI governance should focus on products, sectors, and consequences

- How the post-ChatGPT boom changed AI research and graduate training

- Why access to a chatbot is not the same as democratic control

- What agency and representation would add to AI governance

- Why government and education are struggling to match AI’s pace

- How community-led action could fill the institutional gap

Chapters:

00:00 The one research idea Nidhi would keep

01:50 Privacy, fairness, robustness, and trust

04:54 The unavoidable trade-offs in trustworthy AI

07:54 What Canada gets wrong about AI policy

10:19 Why AI governance should regulate products and sectors

15:11 How the ChatGPT boom changed AI research

22:51 Why Nidhi rejects “democratized AI”

24:39 What democratic AI would actually require

26:18 Should AI become a national political question?

31:34 Building national AI literacy

35:32 Why institutions cannot match AI’s pace

38:54 Community action when government falls behind

41:09 The missing intervention

42:07 What gives Nidhi hope

43:29 Where to follow Nidhi’s work

Connect with Nidhi:

Amii profile: https://www.amii.ca/people/nidhi-hegde

University of Alberta: https://apps.ualberta.ca/directory/person/nidhih

Research website: https://sites.google.com/view/nidhihegde

Podcast:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

Connect with Ayush:

Website: https://ayushprakash.com

LinkedIn: https://www.linkedin.com/in/prakash-ayush/

Instagram: https://instagram.com/ayushprakashofficial

Book:

AI for Gen Z: https://www.amazon.com/dp/0981182135

Fenwick McKelvey joins the podcast to explain why Canada’s AI problem runs deeper than research funding or regulation. Canada helped sustain the ideas behind modern AI and launched an early national strategy, yet it still struggles to turn that history into independent industry, enforceable public accountability, and a clear position between the United States, China, and the European Union.

Fenwick breaks down the difference between regulation, policy, and governance; how hype shapes investment and news coverage; why a small number of companies already make rules for the public; and why strengthening existing institutions may matter more than building another AI regulator. We also discuss ChatGPT in higher education, open models, proactive regulation, political privacy, and the technology he is watching after AI.

Fenwick McKelvey is an associate professor in Communication Studies at Concordia University. His research covers digital politics, internet policy, algorithmic media, and AI governance. His book *SimPolitics: America’s Quest to Solve Politics with Computers* was published by the MIT Press in 2026.

Chapters:

00:00 Why Parliament testimony rarely becomes policy

01:28 Regulation, policy, and governance explained

04:38 Governing AI when nobody knows its real impact

08:04 Should governments regulate AI hype?

11:06 Did Canada build AI and lose the industry?

14:14 Regulation is not Canada’s innovation problem

17:00 Where Canada fits between the US, China, and EU

20:31 Fenwick’s AI governance wish list

25:01 Is ChatGPT ruining higher education?

27:07 How students actually use AI

30:10 Can regulation get ahead of technology?

32:35 The next disruption after AI: quantum

34:44 SimPolitics and where to follow Fenwick

Connect with Fenwick:

Website: https://www.fenwickmckelvey.com/

Concordia profile: https://www.concordia.ca/faculty/fenwick-mckelvey.html

SimPolitics: https://mitpress.mit.edu/9780262053198/simpolitics/

Podcast Info:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

Connect:

Website: https://ayushprakash.com

LinkedIn: https://www.linkedin.com/in/prakash-ayush/

Instagram: https://instagram.com/ayushprakashofficial

Books:

AI for Gen Z: https://www.amazon.com/dp/0981182135

What happens when the AI tools you rely on stop being cheap? Matthew Guzdial argues that becoming dependent on today’s generative AI could leave people and businesses exposed if prices rise or providers disappear.

Matthew is an associate professor of Computing Science at the University of Alberta. We discuss AI’s role in game design, why players object to generative AI, and why synthetic data can’t supply missing ground truth in his research. He explains his prediction of an AI bubble and what he thinks could survive it.

The conversation also covers VR’s adoption problems, Roblox and child safety, platform accountability, and how to talk to children about screens. We finish with his advice for young people trying to decide what to do with their lives. His market forecasts and parenting observations are his views, not established outcomes or clinical guidance.

Chapters:

00:00 Introduction: what makes a good game?
02:12 AI in games before ChatGPT
03:42 Concept art, coding and generative AI
05:48 Why players push back against AI
08:27 The risk of relying on cheap AI
11:41 What could follow an AI bubble?
16:10 Small datasets and synthetic data
17:28 AI winters and the road to AGI
19:08 Escapism, Roblox and child safety
21:04 Why VR struggles with adoption
23:17 How marketing shapes technology expectations
24:53 Should children be banned from games?
28:53 Parenting, screens and autonomy
34:24 Advice for young people
35:47 What gives Matthew hope
36:35 Where to follow Matthew and his games

Connect with Matthew:

Website: https://guzdial.com/
University profile: https://apps.ualberta.ca/directory/person/guzdial
Bluesky: https://bsky.app/profile/matthewguz.bsky.social
Google Scholar: https://scholar.google.com/citations?user=jKqmTbIAAAAJ

Podcast Info:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG
Apple Podcasts: https://podcasts.apple.com/ca/podcast/ayush-prakash-podcast/id1557703631

Connect:

Website: https://ayushprakash.com
LinkedIn: https://www.linkedin.com/in/prakash-ayush/
Instagram: https://instagram.com/ayushprakashofficial

Books:

AI for Gen Z: https://www.amazon.com/dp/0981182135

What if the first honest response to ecological collapse is admitting that we may not know how to fix it? Writer and Dark Mountain Project co-director Nick Hunt explains why climate change, mass extinction and cultural alienation cannot be reduced to engineering problems, and why accepting loss does not have to mean giving up.

We discuss the disappearance of the Aral Sea, the surprising life emerging from its exposed seabed and the difficulty of forcing every story into hope or despair. Nick reflects on humanity’s capacity for destruction, the possibility of leaving good traces and why he turns to art to create meaning in a damaged world.

The conversation then turns to writing, artificial intelligence, travel and the value of getting lost. Nick explains why human-made art still matters, why he writes to think instead of delivering a final truth and what different cultures can teach us about purpose, time and simply being. We finish with speculative fiction, hope beyond hope and the message he would give his future self.

Chapters:
00:00 What is the Dark Mountain Project?
03:55 Why technology cannot solve every crisis
08:09 What remains after the Aral Sea disappears
14:04 Hope, despair and humanity’s destructive side
18:03 Why Nick turns to art instead of destruction
19:50 What is the point of writing if we are doomed?
24:36 Writing what the world may not be ready for
26:06 Travel, writing and the value of getting lost
27:52 Why modern life cannot tolerate aimlessness
31:43 Presence, creativity and the pressure to succeed
33:21 Hope beyond hope
36:26 Why there is no magic-wand solution
38:02 What Nick would tell his future self
39:14 Where to follow Nick Hunt

Connect with Nick:

Website: https://nickhuntscrutiny.com

Dark Mountain: https://dark-mountain.net

Podcast Info: 

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

Connect: 

Website: https://ayushprakash.com 

LinkedIn: https://www.linkedin.com/in/prakash-ayush/ 

Instagram: instagram.com/ayushprakashofficial

Books: 

AI for Gen Z: https://www.amazon.com/dp/0981182135

Adam Dickinson joins the podcast to break down what intuition actually is, why logic alone is one of the most limiting and overlooked barriers to human potential, and what it means that most of us are operating from only a fraction of our true consciousness. We also get into the spiritual frameworks that can expand the way you think and feel, and what Adam believes it will take for humanity to evolve beyond the mind and into a deeper intelligence.

Connect with Adam: 

LinkedIn: https://www.linkedin.com/in/adam-r-dickinson/

Podcast Info: 

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

Connect: 

Website: https://ayushprakash.com 

LinkedIn: https://www.linkedin.com/in/prakash-ayush/ 

Instagram: instagram.com/ayushprakashofficial

Books: 

AI for Gen Z: https://www.amazon.com/dp/0981182135

David Roberts of Mara Labs joins the podcast to break down what microplastics are, why air is one of the biggest and most overlooked sources of exposure, and what it means that most of us are consuming the equivalent of a credit card's worth of plastic every week.

We also get into concrete lifestyle changes you can make right now to reduce your intake, and what Mara Labs is building to tackle the problem at scale.

Connect with David:

LinkedIn: https://www.linkedin.com/in/david-roberts-7170416/

Website: https://mara-labs.com

Podcast Info:

Spotify: https://open.spotify.com/show/1ILhje5HSua1FEOlTyFAhG

Apple Podcasts: https://apple.co/3qXL37W

Connect:

Website: https://ayushprakash.com

LinkedIn: https://www.linkedin.com/in/prakash-ayush/

Instagram: instagram.com/ayushprakashofficial

Books: AI for Gen Z: https://www.amazon.com/dp/0981182135