ISTE 2024 - DENVER, CO - Adobe for Education - Brian Johnsrud - Content Credentials Creating Transparency and Responsibility with Generative AI in the Classroom

·26 min·2 clips

Adobe's Content Authenticity Initiative has 2,000 members creating a 'nutritional label' for digital media to show if AI was involved.

This episode is an interview recorded at ISTE 2024 with Brian Johnsrud, Adobe's Director of Education, Learning, Advocacy, and Thought Leadership. The host introduces Johnsrud, who has appeared on the show before. Johnsrud frames the discussion around authenticity, content credentials, and knowing who created media or if AI was involved. He traces attribution back to ancient manuscripts and academic citation practices. Johnsrud explains the rise of digital media and generative AI has made this more critical, citing fake news and misinformation impacts. He argues every educator, from band directors to librarians, should teach media and AI literacy. Johnsrud describes Adobe founding the Content Authenticity Initiative in 2019 with partners like The New York Times. The coalition now includes over 2,000 organizations across hardware, software, governments, and nonprofits. They developed a universal metadata standard that embeds information like creation date, tools used, and AI involvement. Johnsrud compares this to a 'nutritional label' for media. He suggests teachers can use tools with these features to spark conversations about transparency and ethical responsibility. The host asks if the goal is to make attribution automatically visible online. Johnsrud agrees, noting current metadata is often inaccurate or missing. He envisions news outlets and social media surfacing content credentials easily. Johnsrud discusses AI in classrooms, moving from bans to requiring disclosure. He proposes tools could automatically show teachers how students used AI, like for brainstorming versus writing. The conversation shifts to trust, with Johnsrud emphasizing building classroom trust from pre-K. He connects this to adult citizenship and trusting politicians or media. The host raises AI checkers in higher education, noting their failures. Johnsrud suggests students should also ask if teachers used AI for assessment. He shares Stanford's academic integrity policy where professors cannot monitor exams, which reduced cheating. Johnsrud relates this to his jazz band, where trust is essential. The host critiques AI companies for training models without permission or attribution. Johnsrud explains Adobe Firefly was trained only on Adobe Stock, licensed content, and public domain material to ensure commercial safety. He gives an example where Firefly wouldn't generate Snoopy due to copyright. The discussion touches on copyright limits and training students to avoid legal issues. Johnsrud references Harold Bloom's 'anxiety of influence' theory about writers fearing their work isn't original. He concludes that remixing existing ideas is key to creativity, and attribution matters. The episode ends with resources: contentauthenticity.org for the coalition and adobeforeducation.com for Adobe's education work.

As heard by us

A clear conference-floor talk about trust, AI, and the classroom habits that support transparency.

The piece follows Brian John Jurd of Adobe as he talks through authenticity, content credentials, and the harder question of who made a piece of media and whether AI had a hand in it.

Read the full review in PlayNext →

Why you'd press play

You want a sane way to talk about AI, trust, and who made the thing.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from EduTechGuys - Conference Coverage