In this episode of ChatEDU, Matt and Liz interview Charli Hughes, a recent high school graduate heading to the University of Connecticut in the fall. Charli discusses her summer internship at Medtronic (where she works on the Hugo Robot Assisted Surgery System), and her journey into computer science through robotics competitions.

Charli is one of the most thoughtful voices out there on AI, it just so happens that she is a recent high school graduate. Don’t miss her insights and opinions on what’s next for AI in K-12 education.

In this episode of ChatEDU (Bottom-Up AI: Where Proactive Educators Lead), Matt and Liz explore how educators at all levels can advance their work with AI, regardless of what state or federal policies are in place where they live.

In the News:

• AI as Moral Expert - Matt discusses a recent study comparing GPT-4's moral advice to that of professional ethicists, finding that the AI often outperforms humans in areas like moral soundness and thoughtfulness. Matt and Liz put AI to the test on a juicy moral dilemma. The Skills21 prompt library mentioned by Matt in this section can be found at www.skills21.org/prompts

• Social Media Warning Labels - Liz talks about the U.S. Surgeon General's call for warning labels on social media, similar to those on cigarettes, and Jonathan Haidt's book "The Anxious Generation" about the impact of phone-based childhoods on mental health. Liz goes on to explore a new study exploring the role of AI in child development.

Beneath the Surface:

Matt and Liz provide an update on AI state guidance, and the imperative/opportunity for educators to lead irrespective of guidance from “above” (fed, state, district, or building level)

• The hosts emphasize the importance of looking within to determine AI priorities rather than waiting for top-down guidance.

• They recommend resources like the Teach AI Toolkit, COSN Readiness Checklist, and Michigan Virtual's planning framework for schools implementing AI.

• Recent state level reports include - UT, AZ, NC, OH, and NJ.

Short Stack Chat Back

Claude (via ElevenLabs) provides a Ben Franklin-style response on the pros and cons of using large language models for ethical advice.

Have a question or comment? Please email Matt and Liz at chatedu@edadvance.org

In this episode of ChatEDU (Dear AI, Slow the Heck Down), Matt and Liz discuss the disruptive and problematic consequences of large tech companies moving too fast in the AI space. They unpack various incidents, and the implications of rapid AI advancements on education.

First in Our Intro

Matt and Liz discuss the emergence of non-human AI models, including meerkats, whales, dogs, and elephants. They humorously explore the idea of translating animal noises into “language” and the challenges it poses, particularly with Liz's silent dog, Aloe.

Then in the News:

Voice Mode Updates - Matt provides an update on the much-anticipated ChatGPT 4.0 voice mode. Together, Matt and Liz poke fun at the delay, but also explore how this advanced voice mode might be used in schools.

OpenAI's Library - Liz talks about Sam Altman's focus on creating an old-timey library at OpenAI's headquarters, blending high-tech AI work with a Victorian academic feel. Liz goes on to explore the relevance of libraries in modern education and AI development.

Beneath the Surface:

Matt and Liz discuss the consequences of the "move fast and break things" ethos in the tech industry. They review several AI-related incidents, including, but not limited to:

• ChatGPT's Gibberish Incident: A night of bizarre outputs from ChatGPT.
• Google's Gemini Image Issue: Controversial AI-generated images that sparked debates on diversity and historical accuracy.
• Scarlett Johansson Voice Scandal: The controversy around OpenAI's voice persona, Skye, and its unintended resemblance to Johansson's voice.
• Rock Eating and Glue on Pizza: Logical recommendations from Google’s AI search summaries.

Together, the hosts provide some guidance on how educators might navigate these disruptive AI episodes.

Short Stack Chat Back:

Shakespeare-Style - ChatGPT + ElevenLabs provide a three-sentence summary of the consequences of AI companies moving too fast (in the style of the Bard…the real Bard not the Bard turned Gemini).

Have a question or comment? Please email Matt and Liz at chatedu@edadvance.org

In this episode of ChatEDU (Teens on Gen AI: A New Study from Common Sense Media), Matt and Liz dive into a new report from Common Sense Media that explores how teens are using and perceiving generative AI. The episode provides a comprehensive overview of the latest research on this topic, shedding light on both the excitement and concerns surrounding AI among young people.

First in Our Intro

• Matt and Liz talk about interactive prompting and narcissism :). Check out www.skills21.org/ai (Prompts > Counselor > Role-Playing a Challenging Work Situation)

Then in the News:

• Cara: An AI-Free Space for Artists - Matt introduces Cara, a social media portfolio app created by photographer Jenga Zhang. Cara aims to protect artists' rights by providing an AI-free environment where they can share their work without the risk of it being scraped by AI models.

• Tribeca Film Festival - The festival features five short films created using OpenAI's Sora, a text-to-video tool. This marks the first time Sora is showcased at a major arts festival, highlighting the potential of generative AI in filmmaking. Matt and Liz discuss how schools will need to address each artistic field individually when it comes to generative AI.

• Illuminate by Google - Liz discusses Illuminate.withgoogle.com, an experimental tool that transforms dense academic papers into accessible NPR-style discussions. This tool aims to make academic research more engaging and easier to understand, while addressing issues like hallucinations in AI-generated content.

Short Stack Chat Back:

• Dumbledore-Style - ChatGPT + ElevenLabs provide a whimsical, yet insightful, three-sentence summary of the Common Sense Media report in the style of Dumbledore, capturing the essence of teens' perspectives on generative AI.

Have a question or comment? Please email Matt and Liz at chatedu@edadvance.org

In this episode of ChatEDU (AI Surrender - Professors Throw in the Towel on Writing?) Matt and Liz discuss the controversial notion of AI taking over writing tasks. The episode deep dive is informed by the work of two professors from Carnegie Mellon University who are saying the quiet part out loud, with an AI tool aimed at assisting students with the writing process.

Fist in the news:

1. Mechanistic Interpretability - Matt explains the concept of mechanistic interpretability, likening it to an MRI for language models. This technique, championed by AI safety-focused company Anthropic, helps unpack the black box nature of AI models to understand how they generate their outputs and address issues like bias and hallucinations.

2. AI Tutors in Education - Liz revisits the topic of AI tutors, emphasizing the benefits they bring to students (particularly in math and science). These AI tools can serve as accessible and affordable tutors, helping students understand complex problems step-by-step as a means to increase educational equity.

Short Stack Chat Back - The episode concludes with an AI supported take on the Beneath the Surface topic.

Have a question or comment? Please email Matt and Liz at chatedu@edadvance.org

In this first “Short Stack” episode of ChatEDU (AI Hope or Hype - A K-12 Reality Check) Liz and Matt unpack an important New York Times opinion piece titled "Press Pause on the Silicon Valley Hype Machine” by Julia Angwin. They explore key assertions in the article including:

1. Overstated Achievements: AI's claimed capabilities versus real performance.

2. Unreliable Performance: The effort required to correct AI output versus manual tasks.

3. Limited Improvement Potential: Constraints on AI improvements due to data and energy consumption.

4. Resource Allocation: The diversion of resources from practical innovations to AI developments.

In each area, Matt and Liz discuss how these points resonate in the K-12 context, emphasizing the need for balanced AI literacy and thoughtful resource allocation. They end with their own priority list of K-12 AI in education challenges.

First, in the News -

• Holographics and Wearables: Matt shares his excitement about the new Meta Ray-Ban glasses as well as lightweight AR/AI glasses being developed by Stanford researchers

• AI and Gender: Liz discusses the economic impacts of AI on women, highlighting bias in AI models and the underrepresentation of women in AI labs.

Short Stack Chat Back - The episode concludes with an AI supported take on the Beneath the Surface topic.

Have a question or comment? Please email Matt and Liz at chatedu@edadvance.org

In this episode of ChatEDU (That’s a Wrap - Exploring the Fate of Custom AI Teacher Tools), hosts Matt Mervis and Liz Radday think through the implications of the ChatGPT 4o release, and what it may mean for custom AI teacher tools and the broader landscape AI + education landscape.

First, In the News:

1. Microsoft Work Trends Report

2. • Matt unpacks a new report from Microsoft highlighting the need for AI literacy in schools.

1. AI and Solopreneurs

2. • Liz examines the rise of "solopreneurs," entrepreneurs using AI to operate alone, and draws connections to Capstone projects.

1. NSFW Content and AI

2. • Matt and Liz discuss OpenAI's new “Model Spec” including the possibility of NSFW content in AI outputs down the line.

1. Homework Assistance Apps

2. • Liz highlights the prevalence of AI homework assistance apps in the iOS App Store and the challenges they pose for teachers.

Beneath the Surface Links

• GPT 4-o Math Problems

• GPT 4-o Rocky Interview

• GPT 4-o Real Time Translation

Chat Back: Listener Question

• Matt and Liz address a listener's question about whether schools should appoint a primary AI lead.

In this milestone 10th episode of ChatEDU (AI for Good: The Literacy Component We Might Be Missing) hosts Matt Mervis and Liz Radday explore how US attitudes towards AI impacts our perspective and potentially narrows how we define AI literacy for students.

First in the news:

Segment 1: AI's Role in Reading

• Matt discusses the less highlighted role of AI in reading compared to writing, featuring insights from Marc Watkins who addresses the potential risks AI poses to developing deep reading and analytical skills.

Segment 2: Labor in Developing Nations and AI

• Liz examines the significant yet overlooked contribution of labor from developing nations in shaping generative AI outputs, highlighting ethical considerations and impacts on language, with a focus on the overuse of the word "delve."

Segment 3: Google's AI Tool for English Language Learners

• Matt introduces Google's new generative AI tool designed to assist English language learners through interactive and scenario-based learning, enhancing their practical communication skills.

Segment 4: Generative AI and Trust in Educational Settings

• Liz unpacks a study concerning how generative AI influences trust in student-teacher relationships, discussing the challenges and the evolving expectations that both parties face in an AI-integrated educational environment.

Chat Back: Listener Question

The episode concludes with a listener's question about how to best explain AI to high school students, stressing the importance of understanding AI beyond just generative models and incorporating real-world applications that students encounter daily.