We are reckoning with truths that are a bit uncomfortable. AI is not just a tool, but something people are relating to - personally. For kids and teens, we’re finding that that can be extremely dangerous.

In this episode, Shannon Peavey speaks with Tracy Pizzo Frey, a veteran technology leader including 11 years at Google where she oversaw Responsible AI for Google Cloud. As a consultant, she worked with Common Sense Media to develop a system for assessing the risks of large language models and AGI, with a particular focus on kids and teens.

Tracy brings deep technical experience and knowledge of how technology shapes behavior. Drawing on lessons from social media and emerging research, she explores what feels different about AI.

For young people who are still developing social and emotional skills, these interactions may have unique implications. AI systems are responsive and engaging, but they do not challenge users or help them navigate real-world complexity in the same way humans do. Over time, that difference may influence how teens build coping skills, relationships, and a sense of self.

Through her work with Common Sense Media, Tracy has evaluated leading AI systems and reached some important conclusions. Today’s models are not designed to serve as mental health companions for kids or teens, even though many young people are already engaging with them in ways that resemble emotional support.

Tracy shares how these assessments were created, what they measure, and what they reveal about the current state of AI safety. She offers a grounded perspective on building & using these technologies responsibly, especially when younger users are already deeply involved..

00:20 AI’s effect on kids

02:07 Why harms are specific to kids & teens

04:50 AI is not a search engine

07:27 Kids & teachers are often earliest adopters

08:06 Tech companies know more than they let on

09:15 Common Sense Media’s risk assessment project

09:57 Let’s not repeat mistakes

11:24 Tracy's involvement

13:11 Set your charter

14:50 Bring diverse, multi-disciplinary teams

16:25 Why psychological safety is important

17:38 Distilling masses of information into risk assessments

18:48 Why hype matters

20:30 How the team looked at social media

23:20 Early assessment of potential harms

25:10 Character.ai as precursor to interaction with LLMs

26:00 ‘Everything in the whole wide world’

27:13 Why kids are different

32:18 The danger of so-called frictionless relationships

33:02 The best way to test

36:50 Some surprising findings

39:08 How tech can reshape a worldview

41:02 There are good people, but - business models

43:30 Know the tradeoffs

45:07 The fact-to-fiction scale

46:30 Some positivity

48:00 Books, lawsuits, and resources

📚Resources:

• Common Sense Media (CSM) AI Overview https://www.commonsense.org/ai

• CSM’s Parents’ Guide https://www.commonsensemedia.org/articles/parents-ultimate-guide-to-generative-ai

• CSM’s Parents’ guide to AI Companions & Relationships https://www.commonsensemedia.org/articles/parents-ultimate-guide-to-ai-companions-and-relationships

• Google’s AI Principles https://ai.google/principles/

• Social Media’s Harmful Effects on Children https://www.yalemedicine.org/news/social-media-teen-mental-health-a-parents-guide

• AI & Everything in the Whole Wide World benchmark study https://sites.rutgers.edu/critical-ai/wp-content/uploads/sites/586/2022/01/Bender_AI-and-Everything-in-the-Whole-Wide-World-Benchmark.pdf

• Shannon Vallor’s research https://edwebprofiles.ed.ac.uk/profile/shannon-vallor

• Empire of AI by Karen Hao https://www.penguinrandomhouse.com/books/743569/empire-of-ai-by-karen-hao/

Where to find Tracy:

• Tracy Pizzo Frey, Restorative AI https://www.linkedin.com/in/tracy-frey/

Where to find our host:

• Shannon Peavey https://www.linkedin.com/in/spmad/

🙋🏻‍♀️Where to find Women in Product:

• On LinkedIn https://www.linkedin.com/company/women-in-product/posts/ 

• Website https://womenpm.org/

• Join the Community https://womenpm.org/wip-community/