The origin story of Cognitive Resonance
How the commercial deployment of generative AI led to an unanticipated opportunity
An early “concept sketch” for Cognitive Resonance
So why am I launching Cognitive Resonance?
Two years ago, when I left the nonprofit education organization I founded, I felt the biggest contribution I'd personally made to improving education involved spreading ideas from cognitive science into teaching practice. I believed then and now that understanding how our minds work is important and interesting, and that fostering knowledge in this area will help improve broad swaths of society -- in education, certainly, but also in any professional domain that involves reasoning and complex decisions (which is to say, all of them).
My hope was that in the next phase of my career I could continue to foster understanding of human cognition. My immediate challenge, however, was figuring out a "hook" to get people to share my interest in this. Cognitive science is a bit of a loaded term, one that sounds wonky and abstract, and perhaps not all that relevant to practical use-cases. How could I convince people that understanding how we humans think and learn is critical to improving our overall well being?
And then, boom. Generative AI entered all our lives.
I don't know about you, but the first time I sat down and played with ChatGPT I was simultaneously blown away, confused, and frightened. Blown away because I couldn't believe how adept this new technology was at communicating with me. Confused because it seemed far beyond what I thought possible (I'd been keeping a loose eye on "deep learning" developments in the field for years). And frightened because I didn't understand how it was doing what it was doing, or what it might be capable of doing in the near future.
So, in the spirit of life-long learning, I spent a great deal of time learning about large-language and diffusion models, transformer architecture, autoregression, neural networks, and so on. I did this by reading a LOT of research papers but also, and perhaps more importantly, connecting with many scientists doing incredibly interesting work in this area.
What I quickly realized is that my background in cognitive science gave me a big leg up in figuring this new technology out -- indeed, generative AI emerged from this discipline. What's more, as I started to build my own "mental model" of generative AI, I felt myself...relaxing...a great deal. These tools are complex, no doubt, but they are just tools -- and they are intelligible. Once I started understanding how they work, I worried much less about "extinction risk" or the vague threat that they will destroy the "knowledge economy" as we know it. Understanding cognitive science helped demystify generative AI, and helped me to recognize its strengths and limitations.
Which is not to say there aren't risks! In fact, there are many. One big risk that I'm personally attuned to is the deployment of this technology in education by people who are promising things that generative AI cannot presently deliver. Ten years ago, there was a massive push to "personalize learning" in schools, an effort I resisted vigorously and loudly because I believed (and believe) it runs counter to basic principles cognitive science. Many of those efforts have since died out, but they are being revived under the banner of individualized chatbot tutors. Is this a good idea? Is technology finally capable of delivering this ed-tech dream?
Perhaps you can guess my view (hint: no). But my goal with Cognitive Resonance is not to convince people to share my predictions about the future. Instead, I simply want to help people think critically about this technology. I want to build broad understanding of what makes human minds special and amazing, while critically examining this new tool that of course is ultimate the product of human minds and human ingenuity.