The loudest voices about AI occupy the extremes. On one side, doom: the machines will take everything. On the other, hype: the machines will fix everything, any minute now. Both positions share a convenient feature — neither requires you to do anything. Despair excuses you from learning; hype promises the learning won't be necessary. This site exists for the third position: that AI is a genuinely transformative technology whose benefits flow to the people and organizations that engage with it deliberately, ethically, and with their eyes open. We call that position optimistically pragmatic.
§01The case for optimism
The optimism isn't a mood — it's a reading of the evidence. Field studies keep finding the same shape: when AI tools meet real work, real gains follow, and the biggest gains go to the people still climbing. When 5,179 customer-support agents got an AI assistant, productivity rose about fourteen percent on average — and thirty-four percent for novices. When 453 professionals used AI on writing tasks, time dropped forty percent while quality rose, and the gap between stronger and weaker performers narrowed. AI, used well, is a skill-leveler and an amplifier.
Brynjolfsson, Li & Raymond, "Generative AI at Work," Quarterly Journal of Economics (2025); Noy & Zhang, Science (2023).
Kevin Kelly calls the right disposition protopian: not utopia, but a slow march of incremental betterment — and he argues imagining a desirable future is a duty, because you cannot build what you cannot envision. Dario Amodei's "Machines of Loving Grace" makes the concrete version of that case: safely developed AI could compress decades of progress in health and science into years. That's worth showing up for.
§02The case for pragmatism
And yet. Princeton's Arvind Narayanan and Sayash Kapoor make the most rigorous version of the grounding argument: AI is a transformative but normal technology — like electricity, like the internet. Diffusion is slower than invention. Institutions, habits, and humans mediate everything. The gap between a demo and a dependable practice is measured in years of unglamorous work.
The single most useful study for this philosophy is Harvard Business School's "jagged frontier" experiment. Consultants using AI inside its capability frontier completed more tasks, faster, at higher quality. The same consultants, on tasks just outside the frontier, were nineteen percent more likely to be wrong. AI's abilities don't fail gracefully or predictably — the frontier is jagged, and it is invisible until you've walked it yourself. That finding is the whole argument for learning by doing: no summary, no headline, no vendor deck can hand you the map. Reid Hoffman calls the people who hold both truths "bloomers" — driving forward while tapping the brakes. We just call it being pragmatic.
Dell'Acqua, Mollick, Lakhani et al., "Navigating the Jagged Technological Frontier," HBS Working Paper 24-013 (2023); Narayanan & Kapoor, "AI as Normal Technology," Knight First Amendment Institute (2025); Hoffman & Beato, Superagency (2025).
§03The fundamentals matter more, not less
Here is the remarkable convergence: the people who spent careers defining how good products get built have all looked at AI and reached the same conclusion. Marty Cagan is blunt about it — teams are using AI to produce their crappy roadmaps faster. The technology amplifies whatever process it touches; it makes bad practice faster and good practice compound. Jeff Patton's rule — minimize output, maximize outcome — stops being advice and becomes survival strategy the moment output is nearly free. When anyone can generate endless documents, shared understanding is the moat, because shared documents were never shared understanding.
Alan Cooper warned us decades ago about "dancing bearware" — software so amazing for existing at all that we forgive how badly it dances. A great deal of today's AI experience is exactly that. His answer then is his answer now: design for real people's real goals, and ask the good-ancestor question before shipping — is this the right thing to put into the world? And Jakob Nielsen, the field's great empiricist, calls AI the first new user-interface paradigm in sixty years — users finally state their intent instead of operating commands — while noting in the same breath that the existence of "prompt engineers" is proof the usability isn't there yet. Enormous enthusiasm, blunt honesty about what's broken. That is the posture.
Cagan, svpg.com; Patton, User Story Mapping (O'Reilly, 2014); Cooper, "The Oppenheimer Moment" (Interaction 18); Nielsen, "AI: First New UI Paradigm in 60 Years" (2023).
§04Ethics is a starting point, not a review gate
Most technology curricula bolt ethics on at the end, after the skills are learned and the habits are set. We put the Ethical Crossroads at stage two of six — before the deep skills, right after the spark. This is the framework's most deliberate move, and it has authority behind it: UNESCO's 2024 AI Competency Framework puts a human-centred mindset and the ethics of AI as its first two dimensions, ahead of techniques and system design. MIT's AI literacy curricula interleave ethics with every technical lesson rather than quarantining it. Cooper's "Oppenheimer Moment" explains why: the people who build harmful systems are mostly good people whose intentions were subverted. Good intentions are not a design method. You decide who you're going to be with this technology before it becomes muscle memory — because after, you're only rationalizing.
UNESCO AI Competency Framework for Students (2024); Long & Magerko, "What is AI Literacy?", CHI 2020; MIT RAISE DAILy curriculum.
§05Learning is the strategy
If the frontier is jagged and invisible, then the only durable AI strategy — for a person, a team, or an institution — is a learning practice. Not a tool purchase, not a policy memo, not a conference keynote. A practice. That's what the six stages describe: the spark of curiosity you choose to follow; the ethical crossroads where you set your compass; the labyrinth where difficulty does its quiet work; the rabbit hole where interest becomes yours; the breakthrough where the frontier finally becomes visible; and mastery — fluency maintained by practice, completed only when you turn around and hand the map to someone at stage one.
Ethan Mollick's standing advice is the on-ramp: always invite AI to the table, and stay the human in the loop. And if you lead people, your role in their journey has a name: coach. Optimism supplies the reason to start. Pragmatism supplies the way through. The future belongs to the people who do both.
Mollick, Co-Intelligence (Portfolio, 2024). Full sources for every claim on this site: Research & Influences.
Be optimistic enough to begin. Be pragmatic enough to keep going.
— Chris Tarabochia