Over the past five weeks, we've examined how AI works, where it breaks, and what separates organizations that use it well from those that simply use it often. Most of that has focused outward on tools, models, methodology, and decision-making. This week, we're turning inward. Our Chief Research Scientist Jason Brooks steps back from the data to explore something more fundamental: why smart, capable people consistently underperform when working with AI and what a decades-old finding from cognitive psychology tells us about it. If our earlier posts made the case that prompting is a discipline, Jason explains why it's so hard to practice, even for experts. It's a perspective that reframes the conversation...one we think will stay with you.
Jason writes:
One of the things I do for a living is apply the science of cognitive biases to help clients make better sense of the world and ultimately better decisions about what to do in it. It's an acknowledgement that our brains aren't perfect machines, flawlessly perceiving reality, weighing all possible options with mathematical precision, and choosing the path that optimizes our best interest. No, our animal brains are often faster and messier than that, relying on shortcuts (aka mental heuristics) that help us navigate the complexity of reality with as efficient a use of energy (calories) as possible and as such, there are predictable distortions about how we interpret information, assess risk, and act. Understanding these distortions isn't just academic, it's practical. In life, it can be the difference between frustration and satisfaction in seeking what we want, conflict and camaraderie with the other humans we interact with, and in business, the difference between plans that hit and miss.
If I could only ever know of and teach about just one bias — whether to friends, colleagues, or strangers — it would be the one known as the "curse of knowledge". It's deceptively simple, and like many profound revelations, obvious in hindsight: once we know something, it becomes very difficult to imagine what it's like not to know it and thus we tend to unconsciously assume others have similar knowledge that we do. In 1990, Stanford graduate student Elizabeth Newton demonstrated this with an elegant experiment known as the "tappers and listeners" study. Participants were randomly assigned to be either "tappers" who were instructed to tap out the rhythm of a well-known song (e.g., "happy birthday", "ABC's", etc.) or be "listeners" who have to guess the song being tapped out. Before the listener revealed their guess, the tapper had to predict whether the listener would guess correctly. The tappers predicted listeners would guess correctly about half the time, when in reality, the listeners were only right about 3% of the time. In the tapper's mind, the song was obvious, but to the listener, the taps were just a series of disconnected beats. That gap is an example of the curse of knowledge.
So here you are, a professional in the world of marketing, business, and research, reading a blog like this, which likely means you're pretty smart! But it's precisely because of our intelligence and specialized knowledge that this bias becomes a problem for us. We operate as though others share our context, vocabulary, constraints, and goals. We skip steps in explanation and overestimate our own clarity. In life, that can create friction in relationships, where what feels obvious to us feels opaque or alien to someone else. In business, that can mean misaligned teams, confused or unreached customers, and thus missed opportunities. It's quite easy to forget that the smarter and more experienced we become in a given domain, the more we lose touch with what it's like to be new to it. The great Harvard cognitive psychologist and multiple best-selling popular science author, Steven Pinker, widely known for being a gifted communicator, pretty much dedicated one of his best sellers to this topic in his "The Sense of Style: The Thinking Person's Guide to Writing in the 21st Century", writing "the better you know something, the less you remember about how hard it was to learn. The curse of knowledge is the single best explanation I know of why good people write bad prose."
The curse of knowledge doesn't just affect how we communicate with other people; it also shapes how we communicate with the machines of our time. When we use large language models like ChatGPT, Claude, Gemini, Co-Pilot, Perplexity or Grok for work, we often fall into the same trap. We give vague prompts, incomplete context, and fail to be explicit about our goals, then get frustrated and burn up time when the output falls short (or worse, move forward with slop). But should we be surprised? Does the model know what we're trying to accomplish? Does it understand the constraints of our business and decision-making environment? Does it know the audience, tone, and trade-offs we care about? When we fail to articulate these things, we're effectively tapping out a rhythm and expecting the system to hear the right song.
Here's the twist: as AI technology becomes more powerful and accessible, the skills required to use it effectively become more human rather than more technical. You don't need to write code; you need to write clearly. You don't need to think in algorithms; you need to think in natural language. You don't need to understand what the computer is optimizing for; you need to be clear about what YOU are optimizing for. And that broadens the center of gravity beyond the STEM fields (science, technology, engineering, and math) that built this technology, toward fields built around understanding context, narrative, interpretation, and inquiry. Fields like history, journalism, English and anthropology. These are the disciplines that train people to ask better questions, to surface assumptions, to understand and empathize with their audiences, and to communicate with clarity, even beauty. In the realm of natural language interfaces, those skills are no longer merely "soft"; they're operationally essential.
And worth highlighting, these disciplines have long had built-in defenses against the curse of knowledge. Good journalists not only know how to ask good questions, probing the knowledge of their subjects, but they also assume their readers know less than they do and write accordingly. Good historians know how to reconstruct context from clues without projecting their own modern assumptions backward in time. Good anthropologists immerse themselves in unfamiliar perspectives and question their own lens. Good writers revise with the reader in mind, stripping away ambiguity, and filling in the gaps between what they know of their subject and what their reader has been given on the page. These are all practices of making the implicit explicit, slowing down enough and being intentional about bridging the gap between what we know and what others understand. Providing context, defining objectives, specifying constraints, and assuming nothing is obvious are all principles that will help you achieve success whether you're dealing with friends, colleagues, customers, or AI systems.
I hope I've been able to heed my own advice and communicate it clearly to you, dear reader. While I assume because you've made it this far, you must be a pretty smart person surrounded by other smart people doing meaningful, complex work, I won't assume it's obvious to you that your knowledge can create obstacles to achieving your personal and professional goals. Intelligence and expertise can be celebrated, but if we're mindful of the curse of knowledge bias, we can overcome their hidden burdens and realize an even greater potential of their gifts. The curse of knowledge reminds us that communication is not the transfer of information, but the transfer of understanding. In an age increasingly shaped by artificial intelligence, those who can better bridge the gap between what they know and what others understand may possess one of the most valuable skills of all.