The evidence behind our work

Pioneered "desirable difficulties": making learning temporarily harder leads to stronger long-term retention and transfer.

Referenced in: Is AI Making Us Lazy Thinkers?, What You Choose Not to Automate

Cognitive offloading to external tools frees working memory but may alter skill development when relied on excessively.

Referenced in: Is AI Making Us Lazy Thinkers?

Documented the "cognitive miser" principle: the human tendency to minimize mental effort whenever possible.

Referenced in: Is AI Making Us Lazy Thinkers?

The brain consumes approximately 20% of the body’s energy despite constituting only 2% of body weight, making energy conservation a deep biological drive.

Referenced in: Is AI Making Us Lazy Thinkers?

Described two systems of thought: fast intuitive processing (System 1) and slow deliberate reasoning (System 2). AI use defaults to System 1 unless actively resisted.

Referenced in: Is AI Making Us Lazy Thinkers?

GPS dependence weakens spatial navigation abilities, a model for how tool reliance can erode cognitive skills.

Referenced in: Is AI Making Us Lazy Thinkers?

Studied 758 BCG consultants using GPT-4. Those who used it indiscriminately performed 24 percentage points worse on tasks outside the AI’s frontier.

Referenced in: The Allure of Shortcuts in the AI Era, Keeping Up Is the Wrong Goal, Cognitive Impact

AI assistance can improve short-term task performance even when it impairs long-term learning outcomes.

Referenced in: Cognitive Impact

Users who relied on AI in creative collaboration generated fewer of their own ideas, and the effect carried over to unaided work afterward.

Referenced in: Cognitive Impact

AI compresses quality toward a mean: it improves weaker work but flattens stronger work.

Referenced in: Cognitive Impact

Over four months, people who used large language models performed worse than those who did not, at neural, linguistic, and scoring levels.

Referenced in: The Complete Guide to AI Adoption

Documented the “AI Deskilling Paradox”: doctors using AI for colonoscopies became less adept at finding precancerous growths after just three months.

Referenced in: The Complete Guide to AI Adoption

AI tools may weaken critical thinking by encouraging cognitive offloading.

Referenced in: The Complete Guide to AI Adoption

Describes “agency decay,” a process that operates like muscle atrophy as AI takes over cognitive tasks.

Referenced in: The Complete Guide to AI Adoption

Participants aged 17 to 25 showed higher AI usage and greater cognitive offloading, which coincided with lower critical thinking scores.

Referenced in: The Complete Guide to AI Adoption

The “10-20-70 rule”: 10% of AI success depends on algorithms, 20% on technology and data, 70% on people and processes.

Referenced in: The Complete Guide to AI Adoption

95% of generative AI pilots fail to deliver measurable P&L impact.

Referenced in: The Complete Guide to AI Adoption

Nearly 40% of all AI interactions in organizations involve sensitive data, much of it through unofficial channels.

Referenced in: The Complete Guide to AI Adoption

Initial overtrust in AI is disproportionately harder to correct than initial undertrust. First impressions of AI competence stick, even when contradicted by later evidence.

Referenced in: The Complete Guide to AI Adoption

Even one hour of prompt training “soothes prompt anxiety” and increases willingness to engage with AI tools.

Referenced in: Keeping Up Is the Wrong Goal

Surveyed 211 professionals and found four distinct AI attitude clusters (optimistic pragmatists, neutral observers, skeptical realists, enthusiastic advocates), each predicting different prompting behavior.

Referenced in: AI Perception, Prompting Style

Meta-analysis of 106 studies found that human-AI teams often underperform compared to either humans or AI working alone.

Referenced in: AI Perception

Personality shapes AI perception more than the technology itself. Openness correlates with viewing AI functionally; neuroticism produces negative emotional responses.

Referenced in: Struggling with Your Relationship with AI? Maybe It’s You

Conspiracy mentality correlates with negative AI views, suggesting that resistance to AI is sometimes rooted in broader patterns of institutional distrust.

Referenced in: Struggling with Your Relationship with AI? Maybe It’s You

Analyzed 200,000+ human-AI conversations and found users shift from structured, machine-like prompts toward conversational behavior after their first few interactions.

Referenced in: AI Learner, Prompting Style

AI literacy directly predicts prompting strategy quality, and brief training interventions produce large effects (d=0.91).

Referenced in: AI Learner

Surveyed 100+ papers: structured prompting helps on math and symbolic reasoning (+12-14%), but gains on other task types are minimal (+0.7%). Iterative refinement outperforms single-shot commands by 20-40%.

Referenced in: Prompting Style

Experts with well-structured knowledge can use AI to free up bandwidth for more strategic, high-level thinking.

Referenced in: Is AI Making Us Lazy Thinkers?

Identified the “gulf of envisioning”: the gap between what people want from AI and what they can articulate as a prompt.

Referenced in: What AI Reveals About How We Think

Effective AI use demands metacognition: thinking about your own thinking, articulating goals you have never stated, analyzing processes you normally execute on instinct.

Referenced in: What AI Reveals About How We Think

Heavy media multitaskers perform worse on attention and memory tests than their focused peers.

Referenced in: How Ancient Practices Are Countering Our Modern Attention Crisis

Exposure to opposing political views on social media can increase rather than decrease political polarization.

Referenced in: The Bottleneck Is Us

Social media feed algorithms shape content exposure in ways that contribute to political polarization.

Referenced in: The Bottleneck Is Us

Researches dopamine reward cycles and how variable reward schedules (as found in AI tools) can create compulsive engagement patterns.

Referenced in: The Complete Guide to AI Adoption

Documents the impact of social media on teen mental health, a parallel for understanding AI’s effects on cognitive development.

Referenced in: The Bottleneck Is Us

“The AI Vampire”: coined the metaphor for how AI engagement depletes mental energy through constant context-switching and evaluation.

Referenced in: The Complete Guide to AI Adoption

Deliberate living and resistance to distraction. “Our life is frittered away by detail. Simplify, simplify.”

Referenced in: How Ancient Practices Are Countering Our Modern Attention Crisis, Why I Let AI Program My Presentations

Stabilitas: the monastic commitment to staying in one place and giving sustained attention, a counterweight to modern fragmentation.

Referenced in: How Ancient Practices Are Countering Our Modern Attention Crisis

“Attention is the rarest and purest form of generosity.” Attention as moral practice, not just cognitive skill.

Referenced in: What You Choose Not to Automate

“Deep reading is our species’ bridge to insight and novel thought.” Research on how reading builds the neural circuits for sustained thinking.

Referenced in: How Ancient Practices Are Countering Our Modern Attention Crisis

“Do you have the patience to wait till your mud settles and the water is clear?” Stillness as the prerequisite for clarity.

Referenced in: The Bottleneck Is Us

Convivial vs. industrial tools, radical monopoly, and the two thresholds beyond which a tool begins to dominate rather than serve its users.

Referenced in: Why I Let AI Program My Presentations, What You Choose Not to Automate

Antifragility: systems that grow stronger from stress and disorder, and the hidden costs of removing friction from complex processes.

Referenced in: The Allure of Shortcuts in the AI Era

AI’s mastery of language represents a qualitative shift: the first technology that can generate stories, manipulate emotions, and form relationships.

Referenced in: The Bottleneck Is Us

The productivity paradox: transformative technologies take decades to deliver their promised gains because organizations must restructure around them first.

Referenced in: The Wrong Room

Reorganized factories around workflow rather than machine type, demonstrating that the value of a new technology depends on restructuring work around it.

Referenced in: The Wrong Room

Two dystopian visions: Orwell feared imposed control; Huxley feared we would come to love what oppresses us. AI echoes the Huxleyan warning.

Referenced in: Struggling with Your Relationship with AI? Maybe It’s You

Flow: the state of optimal immersion that emerges from matched challenge and skill. AI disrupts flow by collapsing the challenge side of the equation.

Referenced in: The Curious Case of the AI Hangover

Negative capability: the capacity to remain in uncertainty “without any irritable reaching after fact and reason.” A quality AI use tends to erode.

Referenced in: What You Choose Not to Automate

“I write entirely to find out what I’m thinking.” Writing as a tool for discovering thought, not merely expressing it.

Referenced in: The Draft Decision Framework

The Phaedrus dialogue: Socrates argued that writing would weaken memory and create an illusion of understanding. The same argument resurfaces with each new cognitive tool.

Referenced in: Same Machine, Different Courage

“I quote others only in order the better to express myself.” Authorship as synthesis and curation, not pure origination.

Referenced in: Same Machine, Different Courage

Cogito ergo sum: grounding existence in the act of cognition. What remains when the thinking itself is delegated?

Referenced in: The Bottleneck Is Us

Potential space: the zone between inner and outer reality where play, creativity, and authentic expression emerge without judgment.

Referenced in: Same Machine, Different Courage

The Hero’s Journey: the narrative structure of transformation through challenge and growth. AI adoption follows the same arc.

Referenced in: Struggling with Your Relationship with AI? Maybe It’s You

“Two roads diverged in a wood, and I took the one less traveled by.” The choice of effort over convenience as a source of meaning.

Referenced in: The Curious Case of the AI Hangover

Distinguishes apara vidya (lower knowledge, information) from para vidya (higher knowledge, direct experience). AI excels at the first and cannot touch the second.

Referenced in: The Bottleneck Is Us

Etuaptmumk (Two-Eyed Seeing): the Mi’kmaq practice of integrating Indigenous and Western knowledge systems, a model for holding multiple ways of knowing.

Referenced in: The Bottleneck Is Us

“We have Paleolithic emotions, medieval institutions, and godlike technology.” The gap between human development and tool power defines the AI challenge.

Referenced in: The Bottleneck Is Us