429 | AI, Education, and the Decline of Independent Thinking | Jonathan Schaeffer

Are you using AI to make better decisions—or letting a confident answer do your thinking for you?

In this episode of Futureproof Founder, Jeff Mains sits down with AI researcher and entrepreneur Jonathan Schaeffer for a candid conversation about privacy, accuracy, and the gap between what AI can do and how much we should trust it. After 40 years in academia, Jonathan stepped into entrepreneurship to address the problems he saw eroding public trust in the technology he had spent his career advancing.

Jonathan shares the classroom frustrations that inspired Kind, a tool designed to answer questions using a person’s own collection of documents, videos, and other materials. What began as a way to reduce repetitive student emails revealed an unexpected feature users loved: finding the exact moment a topic appeared in a recorded lecture.

Key Takeaways

[~08:00] AI should inform decisions—not make them for you. Jonathan shares his Star Trek–inspired vision of AI: provide useful information, acknowledge uncertainty, and keep humans in control.

[~15:30] A confident answer is not necessarily a correct answer. Jonathan explains why persuasive AI responses still require verification and why he believes “hallucination” softens the seriousness of factual errors.

[~18:30] Solve a problem you understand firsthand. Repetitive student emails and unreliable online study materials inspired Kind—a tool designed to answer questions from a curated collection of course content.

[~22:30] Your customers may discover your product’s strongest selling point. Students especially valued jumping directly to the relevant moment in a recorded lecture—an unexpected benefit that encouraged Jonathan to keep building.

[~30:00] Privacy matters—until convenience gets in the way. Jeff and Jonathan explore the challenge of building a business around a value customers say they prioritize but frequently trade away.

[~37:30] Keep the vision steady while the technology evolves. Jonathan revisits his original 2017 product concept and explains how advances in AI strengthened the implementation without fundamentally changing the vision.

[~40:00] Local computing creates different cost-and-speed tradeoffs. Jonathan describes using otherwise idle desktop resources for substantial analysis—and why he is willing to wait longer rather than incur cloud-processing charges.

[~50:30] Founder conviction is not proof of customer demand. Before shipping, ask whether people will buy what you are building. Jonathan discusses validating a real need through adoption, usage, and willingness to pay.

[~52:30] Give customers a clear choice between local and hybrid processing. Jonathan explains the differences between Kind Local and Kind Hybrid, using handwriting recognition to illustrate the tradeoff between fully local processing and cloud-assisted capability.

Tweetable Quotes

“But let's call it what it is. It's an error. It's a lie. It's a fabrication.” — Jonathan Schaeffer, on AI “hallucinations”

“This goes nowhere without trust.”— Jonathan Schaeffer

“If you're meeting a real need, people will buy the product. And just because I believe there's a need doesn't mean that there is a need.”— Jonathan Schaeffer

“The product that lasts won't be the fastest one. It'll be the one people can actually trust with what matters: their data, their decisions, their thinking.”— Jeff Mains

SaaS Leadership Lessons

Build trust into the product—not just the pitch.

Jonathan’s approach makes privacy and the boundaries of the knowledge base central product decisions. The leadership takeaway: define what your product should access, what it should answer, and what customers must remain in control of before making bigger promises about capability. Pasted text

Pay attention to the feature customers keep celebrating.

Kind began with a goal of reducing repetitive questions, but students highlighted the value of searchable video moments. Customer feedback can reveal a more compelling benefit than the one that originally inspired the product. Let those discoveries influence what you improve and how you explain its value. Pasted text

Keep the vision steady while adapting the implementation.

Jonathan’s original concept remained recognizable years later, even though advances in AI dramatically improved how it could work. Separate the enduring customer problem from the technology used to solve it. Your implementation can evolve without abandoning the reason the product exists. Pasted text

Validate principles with customer behavior.

A value such as privacy may matter deeply to you without motivating every buyer. Jonathan acknowledges that distinction directly. Identify the customers for whom your priorities solve a meaningful problem, then look for evidence in usage, feedback, and willingness to pay—not agreement alone. Pasted text

Use AI to expand capability without surrendering accountability.

Jonathan’s argument is not that leaders should avoid AI. It is that they should remain responsible for validating information and deciding how to use it. For SaaS teams, the practical lesson is to preserve critical thinking alongside AI adoption rather than treating persuasive output as a substitute for judgment. Pasted text