Ethan Mollick2026The follow-up to Co-Intelligence
The Next Phase of AI
AI no longer just answers. It takes a task and comes back with the work done. Your part moves to the two ends: the brief and the check.
The core idea
Ethan Mollick is a professor at the Wharton School who studies how AI changes work and education, and he writes the newsletter One Useful Thing. Co-Existence is his follow-up to Co-Intelligence, written about thirty months later, when AI systems had started to take on whole tasks by themselves. He argues that the future of AI will be shaped by how people interact with it, not by simple replacement. His working ideas are the jagged frontier, where AI is strong at some tasks and weak at others in ways that are hard to predict, the capability overhang between what the tools can do and what people use them for, and four human advantages worth building: deep knowledge, wide knowledge, taste and agency. The second half looks at jobs as bundles of tasks, at schools and at society, drawing on his own experiments and on research he has run with companies and colleagues.
The shift is in the title. Co-Intelligence treated AI as something you sit down with: you ask, it answers, you think together. In Co-Existence the systems act. They read for you, run tasks for you and filter what reaches you. The useful question becomes which parts of my work do I hand over, and how do I keep my judgement sharp while I do?
Mollick does not claim to know where this ends. He keeps his tests running, says plainly what still puzzles him, and puts his weight on the things a person can build: knowledge deep enough to judge the output, wide enough to know what to ask for, the taste to choose, and the agency to try.
Co-Intelligence · 2024
You prompt, read, push back and stay in every step. The centaur and the cyborg.
about thirty months later
Co-Existence · 2026
Agents take whole tasks and return finished work. You brief, then judge. The conjurer.
AI is strong at some tasks and weak at others that look easier, and the line is hard to see from outside. You find it by testing your own work, and you test again when the models change.
What the models can do has run ahead of what people and organisations use them for. The gap is a risk for anyone who ignores it, and an opening for anyone who explores it.
A new way of working beside the centaur and the cyborg. You summon an agent for a task, let it work, and dismiss it, which makes your job look more like a manager's than a maker's.
Interactive · the handoff desk
Pick one real task from this week. Four questions suggest how to work it: hand it off, split it, work it together, or keep it. You get a brief to paste into your AI tool and a list of checks to run before the result goes anywhere.
Your handoff
Answer the four questions to see how to work this task.
The centaur, cyborg and conjurer are the book's. The questions, the rule that picks one, "keep it yours" and the checks are HourLife's. Your answers stay in this browser.
Concept anatomy
Each row is one task from start to finish. The amber lane is you, the cyan lane is AI. As the modes go from centaur to conjurer, your share of the time shrinks, and what is left of it is judgement.
Co-Intelligence
A clean split. You do your part, AI does its part, and you join them.
Co-Intelligence
Interwoven. You and AI take turns on the same piece of work, round after round.
New in Co-Existence
Summon and dismiss. You set the task, an agent does all of it, and you judge what comes back.
YouAIHourLife's drawing of the three modes
The book's answer to what a person should build. Two are about knowing, one about choosing, one about acting. The "build it" lines are HourLife's.
Six chapters on how AI and people think and work, two codas on what follows, and a return to the title.
The argument
The First Coda
The Second Coda
And to close
Community insights
Nº 01HourLife takeaway
In Co-Intelligence, AI was a partner you sat down and thought with. Here Mollick describes systems that act on their own: they run tasks, read and filter information, and come back with finished work. The title marks that change, from thinking together to sharing the same space.
Nº 02HourLife takeaway
This is the jagged frontier, carried over from the first book. In a field experiment Mollick co-ran with Boston Consulting Group, consultants using AI did better on tasks inside its abilities and worse on a task outside them, and they could not reliably tell which was which. The edge also moves with every new model, so an old verdict goes stale.
Nº 03HourLife takeaway
Mollick calls this gap the capability overhang. Models have improved faster than our habits, our jobs and our organisations have changed around them. The gap is also an opening for anyone willing to explore it.
Nº 04HourLife takeaway
The book adds a third way of working to the centaur and the cyborg from Co-Intelligence: the conjurer. You set the direction, summon an agent to do the work, then judge what comes back. It is closer to managing than to doing, and it fails when nobody sets a clear goal or checks the result.
Nº 05HourLife takeaway
Mollick looks at work task by task instead of asking whether a whole job survives. Some tasks go to the machine, some stay with you, and the mix changes what the job is. Seeing your own work this way makes the change easier to plan for.
Nº 06HourLife takeaway
These are the first two of the book's four human advantages: deep knowledge and wide knowledge. Deep knowledge lets you direct the work and catch the errors. Wide knowledge lets you see where AI could help in the first place, because you cannot ask for what you have never heard of.
Nº 07HourLife takeaway
The third advantage is taste: knowing what good looks like and why one draft beats another. When generating options becomes cheap, choosing well between them becomes the scarce part of the work.
Nº 08HourLife takeaway
The fourth advantage is agency, in Mollick's sense: the nerve to try things, explore what the tools make possible and act on it before anyone tells you to. It ties back to the capability overhang. Somebody has to go first.
Nº 09HourLife takeaway
The book warns against leaning too hard on systems that are persuasive and good at imitation. Treat a confident reply as a draft to check, and keep doing enough of the work yourself that you can still tell the difference.
Action steps
Pick three real tasks from this week, one easy, one medium and one hard. Give each to the best AI model you can use and write down where it was strong, where it was weak and where it was wrong.
List 15 to 20 tasks from a normal week. Mark each one: AI could do it end to end, AI could draft it for you, or it needs you. Circle the two tasks you most want to keep.
Write half a page an AI can read before it works for you: your role, who the work is for, what good looks like, and what it must never do. Save it where you can paste it in.
Before you delegate the next task, write down the two or three things you will verify before the result goes anywhere: a figure, a source, a name. Check them every time.
Write one line each for your deep knowledge, wide knowledge, taste and agency. Pick the weakest one and choose one way to build it this quarter.
Keep your frontier test tasks in a note. Rerun them whenever you switch tools or a new model comes out, and update what you hand off.
Spend thirty minutes a week on a long article or chapter from a field you do not work in. It feeds the wide knowledge that tells you what to ask for.
Each week, write the first draft of one task yourself before you ask AI. It keeps the skill you need to judge its output in practice.
Before a decision that matters, ask one colleague or friend who is likely to push back. A model can be prompted to argue, but a real objection carries more weight.
Let the machines take the tasks. Keep the judgement, and keep it in practice.HourLife note after Ethan Mollick
Pick setup actions and daily habits from Co-Existence: The Next Phase of AI, choose a start date, and download a printable checklist with real calendar pages.
Ethan Mollick is a professor at the Wharton School who studies how AI changes work and education, and he writes the newsletter One Useful Thing. Co-Existence is his follow-up to Co-Intelligence, written about thirty months later, when AI systems had started to take on whole tasks by themselves. He argues that the future of AI will be shaped by how people interact with it, not by simple replacement. His working...
HourLife’s key takeaways include: The question has moved from how to use AI to how to live and work beside it. AI is brilliant at some tasks and poor at others that look easier, so you find the edge by testing, not guessing. The tools can already do far more than most people ask of them.
It's a strong pick for readers exploring Build a Career You're Proud Of, Decision Making, and Digital Literacy. HourLife distills its core idea into key insights and practical actions worth trying.
Run your own frontier test — Pick three real tasks from this week, one easy, one medium and one hard. Give each to the best AI model you can use and write down where it was strong, where it was weak and where it was wrong.
About five minutes. The HourLife summary distills Co-Existence: The Next Phase of AI into its core idea, 9 community insights, and 9 practical actions you can apply right away.
One idea from one book and one small habit from another, every week or once a month, on the day and around the hour you pick, where you are. Nothing else, and one link stops it.
Print it, pin it, post it. Ways to take Co-Existence: The Next Phase of AI off the screen and into the world.
The core idea, key takeaways, and how to apply Co-Existence: The Next Phase of AI, as one clean, readable page.
Read summaryEvery action from this page as a printable to-do list, with a 7-day tracker.
Download PDFA shareable card with the book and its top-voted insight, for saving or posting.
The summary card and every quote card in six sizes: Instagram feed, Story, Pinterest, YouTube thumbnail, phone wallpaper, and link preview.
Save the insights that land, build the habits that stick, and keep what you learn from Co-Existence: The Next Phase of AI working long after you close the page — inside HourLife.
