See How AI Is Transforming Education Right Now! (Claude + Gamma Demo)

Full video transcript

Kelly Tenny joins the Scale With AI Summit for a spotlight session on AI in education, hosted by Darby. Kelly is a college professor with a master’s in curriculum and education, has taught in the K-12 public school system for nearly 16 years, and co-created a UC Berkeley course on AI in business with Darby before ChatGPT had even launched.

Kelly opens by framing everyone as an educator in some form, whether running a classroom or teaching clients in a business context. She describes the pressures facing traditional education: budget cuts, larger class sizes, fewer teachers, and a persistent demand for personalized, engaging learning that accounts for different reading levels, learning disabilities, and language backgrounds. She notes that in 1999, when she started teaching, building a lesson that addressed those varied needs meant writing everything from scratch by hand. Historically, building a single slide deck or webinar presentation could take hours or days; many people admit they sometimes skipped it because of the time cost, or paid someone else to do it.

She positions AI as a multiplier rather than just a shortcut: work that used to take two days can now take an hour, and in that hour it’s possible to generate far more content than before, including personalized materials, handouts, and notes for people with different learning styles. She connects this to microlearning research showing people retain content best in 12-to-15-minute chunks, in contrast to traditional 1-to-3-hour webinars, master classes, and college lectures.

Kelly then demonstrates her live workflow. She uploads a PDF of a nonfiction book (not an education text) to an AI tool and asks it to create a course syllabus with a course outline and objectives, as if the book were a textbook for a class she’s designing. The tool returns a full syllabus in minutes, broken down by knowledge, skill-development, and application objectives – a task she says used to take her three to four hours per class when she started teaching college 17 years ago, multiplied across the five classes she teaches. She and Darby recall that when they built their UC Berkeley course, the syllabus was the first thing requested, and that course was later approved for continuing education credit for Florida realtors.

From the syllabus, Kelly pulls the outline for a single week and asks the AI to turn that section into slide-deck content, keeping it grounded in the original text rather than starting fresh in Gamma. She and Darby compare notes: ChatGPT offered to format the output with dashes that Gamma reads as page breaks, and at one point suggested exporting or uploading directly to a Gamma account – which Darby says is likely a hallucination for now, though Gamma has recently added an API enabling MCP-based upload workflows. With Claude, this auto-formatting didn’t appear explicitly, but Gamma still parsed heading levels (H1, H2, H3) from pasted text to determine slide breaks in either case. Darby gives a brief explainer of MCP (Model Context Protocol) as a way for AI tools and platforms to interoperate, similar to an API, pointing to an earlier summit session by Roger Campbell II for more background.

Kelly then pastes the outline into Gamma using “paste in text,” selects “preserve text” so Gamma keeps her exact wording rather than generating new content, and clicks generate. She explains her preference for exporting the result to PDF or PowerPoint rather than relying on a Gamma share link, citing reliability: PowerPoint and PDF files open on almost any computer, which matters when Wi-Fi fails or a link won’t load at a live teaching event. Darby notes Gamma also integrates with PowerPoint and Google Slides directly.

She extends the demonstration by asking the AI tool to draft a worksheet for students, aimed at those who may struggle with listening, need to take notes, or want to focus on key points. She frames resources like workbooks and handouts as ways to add real value to paid courses and master classes, potentially supporting higher pricing. She also shows that the same source material can be used to sketch out ideas for micro-courses, mini-courses for lead magnets, or scripted video content, all traceable back to the original uploaded text.

Kelly emphasizes the importance of building from a real source text specifically for educational content, distinguishing it from simply asking a chatbot to generate material freely. She connects this to broader concerns about AI-generated content in academia, including a cited case of a Princeton professor facing scrutiny for AI-written teaching slides, and mentions she has taken on a role leading a California State University System committee tasked with creating policy on the ethical use of AI for students and faculty.

The discussion shifts into audience Q&A. Asked how textbooks are changing, Kelly describes the shift from physical textbooks to accessible ebooks already underway in higher education for cost and environmental reasons, and speculates about interactive and AR-based educational materials, referencing her consulting work with an augmented-reality company exploring AR anatomy models for classrooms. On teacher-created ebooks and textbooks, she predicts growth but notes lingering institutional caution around AI-authored material, describing a possible future workflow where a teacher selects specific chapters from multiple existing books and has AI compile them into one custom, non-AI-authored text tailored to a syllabus, a service currently offered commercially by textbook companies.

On a question about the “dead internet theory,” Kelly says she isn’t deeply familiar with the concept but agrees that AI-generated content, comments, and even follow-up comments are increasingly visible on platforms like LinkedIn and Reddit. She expects this to push people toward more intentional communication and trusted, curated sources rather than broad reach for its own sake. She connects this to media literacy in the classroom, recalling how she taught middle schoolers to evaluate source validity using the “tree octopus” hoax website as an example, and suggests that as AI-assisted deep research improves, source evaluation will become an even more central skill.

Kelly also shares that she’s currently using tools like Descript and HeyGen to help a client – an elderly couple, one dealing with dementia – preserve decades of educational content and legacy work, and mentions doing similar legacy-preservation work for client Les Brown. She frames these video and AI tools as freeing people up to focus on relationship-building and one-on-one service rather than manual production work.

The session closes with Darby thanking Kelly and inviting viewers to like, comment, subscribe, and follow for more AI and innovation content, followed by sponsor acknowledgments for Gamma, Pickaxe, Side Hustle, and Market Sauce.