Achieve Peak Creativity: Merging Flow States with AI Technology w/ Steven Kotler | EP #151
Summary
Kotler sees AI as an additive tool for creativity, but leaves the ultimate human–AI comparison unresolved. AI can raise lower performers toward the middle and may lift top creators much higher, though he has no data for the latter. When asked whether a thousand-times-better AI could be as creative as a human, he answers “yes, probably,” then asks whether it would work better alone or with a human.
The creative economy is already a material growth market: Kotler says it tripled over 15 years, with its “super-creative core” expanding from roughly 100 million to 300 million people. He cites about $1 trillion and 4.5% of US GDP, $3 trillion to $3.2 trillion globally, and a 13% earnings premium for workers who use creativity. Whether AI expands or devalues creative work remains an open question.
Flow is presented as a trainable human-performance stack, not an advantage machines automatically erase. Kotler cites a self-reported 500% productivity gain, a reported 700% creativity gain, and 40% to 60% increases across eight creative stages. Flow Research Collective’s 15,000 participants across 160 countries and 28 industries reportedly increased time in flow by 73.8%, often moving from once or twice weekly to twice daily.
Today’s AI excels at convergent pattern recognition—“matching like with like”—while Kotler says humans remain better at lateral thinking, or linking “unlike with unlike.” Kotler himself relays a medical study with roughly 75% accuracy for a physician, 80% for physician-plus-AI, and 95% for AI alone. He agrees that diagnosis is convergent and says it does not settle the question of divergent creative work.
Abundance could undermine its own beneficiaries if automation removes the challenge that sustains flow, meaning, and motivation. Kotler places the productive challenge-skill balance around 4% to 5% beyond current ability. Diamandis invokes the Universe 25 rat experiment, while both consider virtual worlds, “technological socialism,” and “made by humans” signals as possible responses to an overly automatic future.
Neurotechnology makes the human side of the race dynamic. Kotler says his lab identified five or six neural markers for the transition into flow; signals that seemed to require a roughly $500,000 fMRI experiment three years earlier can now be measured with portable EEG. He describes this as preceding the next wave of brain-computer interfaces.
The central AI-safety bottleneck is cooperation at scale during what Kotler expects could be a “bumpy next 20 years.” He compares AI with social media and drug legalization: technologies can be addictive or disruptive when societies are unprepared. Group flow and long-term planning may help, but the transition is the dangerous period.
The operating playbook combines mindset training, correctly timed deep work, immediate feedback, and community. Kotler recommends bed-to-desk in under five minutes, a page of writing per day, brief creative practice, nervous-system regulation, and treating flow as a four-stage cycle. His new eight-month Alliance is designed for about 100 super-creatives, with three live events, training, AI instruction, accountability, and a June launch.
Deep dive
1. Exponential technologies have moved from forecast to operating reality
Diamandis and Kotler published The Future Is Faster Than You Think in 2019, about a month before the pandemic. Since then, AI has become “a thing now—it’s real,” Tesla Full Self-Driving has stopped trying to kill Diamandis, Starship is flying, Bitcoin reached $100,000, and quantum computing has begun to materialize.
Google’s Willow chip is their clearest quantum marker. Kotler says Google quantum chief Hartmut Neven showed that increasing qubit counts brought a reduction in error rates; he calls that “the realization.” The two planned to ask Neven about quantum computing and consciousness.
Kotler cautiously connects falling uncertainty to the free-energy principle, which he describes as a mathematical principle in which systems diminish uncertainty. A paper published that day, he says, extended the principle into intuition and flow; he wonders whether flow could be a foundational property of the universe without claiming that question is settled.
Their next joint book revisits Abundance roughly 14 years later. Working titles are The Age of Abundance and We Are as Gods, with publication discussed for 2026, possibly around Abundance360 in March. Kotler says the book will examine the last-mile problem: basic living standards have risen, but society has not fully recognized or distributed that progress.
2. AI raises creative altitude more reliably than it saves time
Kotler sees creatives approach AI expecting productivity, speed, or labor savings, then abandon it when those promises do not immediately materialize. His counterintuitive diagnosis is that AI often does not make the task faster; it “levels you up,” helping users work at a level they could not normally reach.
Diamandis supplies the model of excellent prose: roughly 90% of a sentence meets expectations, then the final word breaks the pattern, captures attention, and lands the meaning. Large language models tend toward the common or average continuation, although users can adjust their parameters to make outputs more fanciful.
Diamandis believes systems such as Gemini 2.0 and future generations from OpenAI and Claude will eventually analyze Kotler’s corpus and mimic his writing style. Kotler does not reject that possibility.
Kotler says a recent novel finally brought “40 years of practice” together at age 57. He and Diamandis also argue that around age 50, communication between the brain’s hemispheres can increase creativity, intelligence, perspective, empathy, and measurable wisdom. Both say they currently feel more capable than they did when younger.
3. Human lateral thinking is the present edge, not a permanent moat
Kotler’s labor-market framing is asymmetric: AI “raises the bottom to the middle” with little friction, bringing lower-performing employees toward standard. He has not seen research on the top end, but says, “I guarantee you” elite creators are being lifted much higher—a conviction, not an established finding.
AI is already superior at pattern recognition, which Kotler defines as matching “like with like” and discovering close connections. Innovation depends more heavily on lateral thinking—linking “unlike with unlike” across distant conceptual domains—and he argues humans remain better at that because most AI systems are built as convergent rather than divergent thinkers.
Diamandis asks whether an AI system one thousand times better could be as creative as a human in every discipline. Kotler answers “yes, probably,” then shifts the question: would that AI be more creative alone or with a human in the chain?
Kotler relays a medical benchmark of approximately 75% accuracy for a physician alone, 80% for physician-plus-chatbot, and 95% for the chatbot alone. Diamandis responds that this is convergent diagnosis, not divergent thinking, and Kotler agrees that the example does not settle the creativity question.
4. Effortless abundance could extinguish the challenge that produces meaning
Diamandis argues that creativity is fundamental to human health and well-being, so humans are unlikely simply to abandon it when machines become capable. Kotler’s concern is that flow requires a challenge about 4% to 5% beyond present skill; remove difficulty, and automation may remove a mechanism supporting happiness, purpose, meaning, and life satisfaction.
Diamandis’s analogy is Universe 25, which he describes as a 1960s New York experiment beginning with four breeding pairs of rats inside a habitat with plentiful food, nesting space, and territory. He says the population ultimately died out because it had no challenge.
Their speculative resolution is “technological socialism,” where technology rather than the state supplies material needs, while virtual worlds continuously tune difficulty to the user’s sweet spot. Kotler’s relevant singularity is the moment simulation can provide a fuller or more pleasurable human experience than physical reality.
Kotler rejects technological inevitability. Society is already reconsidering social media after learning about its addictive costs, and he expects “AI-free countries” or technology-limited jurisdictions to draw deliberate lines. Diamandis says “made by humans” is already becoming a tag.
5. Creativity is a multitrillion-dollar economy
Drawing on Richard Florida’s categories, Kotler separates professional creatives—doctors, accountants, and other specialists introducing novelty within a field—from the “super-creative core” of creative entrepreneurs, artists, and writers who invent culture and the future. The latter group, he says, grew from about 100 million to 300 million people.
Over roughly 15 years, the broader creative economy tripled. Kotler places it near $1 trillion in the United States, about 4.5% of US GDP, and between $3 trillion and $3.2 trillion globally. Adobe’s 2016 State of Create study reportedly found that people who use creativity in their work out-earn noncreatives by 13%.
The management signal predates generative AI: an IBM study of CEOs in 2010 ranked creativity as the quality most likely to help a CEO thrive in the 21st century. A later study returned analytical thinking, problem-solving, and creativity, but Kotler calls that largely redundant because exceptional analysis and problem-solving are themselves creative.
Diamandis asks whether AI’s ability to bring more people into creative work will de-monetize creativity. Neither speaker claims to know how the market will divide baseline production from higher-order novelty.
6. Flow training is turning a rare state into repeatable infrastructure
Kotler begins with caveated self-reporting. A McKinsey study of CEOs found them 500% more productive in flow; the Flow Genome Project found that people reported being 700% more creative. Because the latter sounded implausibly large, his team separated creativity into eight stages, from problem identification through implementation, and found 40% to 60% increases in each.
Flow Research Collective’s Zero to Dangerous program has trained about 15,000 people in 160 countries and 28 industries. Kotler reports a 73.8% increase in time spent in flow, with many participants moving from once or twice per week to roughly twice per day.
This changes the human-versus-AI comparison because “humans aren’t static in this picture.” The same period producing machine intelligence is also producing meditation apps, portable neurotechnology, and better ways to study and train consciousness.
Kotler’s lab identified five or six neural markers that appear during the transition into flow. Three years earlier, he says, measuring the relevant signals seemed to require a roughly $500,000 fMRI experiment. Through a partnership with VITAL Neuro, a portable EEG headset can now measure the same signal; brain-computer interfaces are the next wave.
7. Human augmentation remains part of the AI comparison
Kotler says there is a revolution in humans’ ability to take advantage of consciousness, beginning with meditation apps and portable neurotechnology and preceding the next wave of brain-computer interfaces. His argument is that forecasts should not increase machine capability while freezing human capability at its historical baseline.
Diamandis imagines a personal AI called Jarvis changing music, modulating difficulty, and presenting challenges that hit flow triggers. Kotler says his team has a dedicated research line on interface design: what is the best way to work with an AI to produce flow?
The ultimate comparison remains unresolved. Diamandis asks when AI will replace a person, or when AI will be better than a person working with AI. Kotler hopes the answer is not within their lifetime but admits he has difficulty believing that it will not happen.
8. Unbounded machine creativity turns safety into a coordination problem
Diamandis quotes Naval Ravikant’s claim that making AI safe may be impossible because “creativity by its nature is unbounded.” Kotler calls it a good point and returns to the broader question of whether technology is an unstoppable force or whether societies can still choose how to engage with it.
Kotler’s “last-mile problem” is that humanity already possesses much of the technology needed to raise global living standards but has not learned to cooperate at scale. He says the only times societies cooperate globally are generally during war or a pandemic; Diamandis adds that pandemic cooperation lasted roughly a month before people splintered.
Group flow—the cooperative form of flow—may offer part of the missing mechanism. Without global conversation and preparation, Kotler fears the generation lost to AI could fare worse than the generation lost to social media, another addictive technology deployed before its effects were understood.
Kotler uses drug policy as a comparison. He describes Portugal as having completely decriminalized everything and says its drug problems continued to fall, while warning that exposing an unprepared generation to highly addictive substances can still be disastrous. Because human brains privilege the present, he expects a “bumpy next 20 years.”
9. Wise AI requires upbringing, pluralism, and restraint
Kotler describes AI as humanity’s progeny: how it is fed, trained, and educated could distinguish a Clark Kent raised into Superman from the same powerful child raised in a destructive environment. Kindness alone is insufficient because kindness can sometimes become weakness.
Diamandis relays Elon Musk’s stated goal of making Grok “maximally curious and maximally truth-seeking.” Kotler initially says that sounds reasonably dangerous, then distinguishes it from maximally creative AI, which he considers more dangerous.
Kotler would rather live a decade from now with digital superintelligence—a billionfold increase in intelligence—than without it. In his best-case scenario, a wise AI could restrain human short-termism and persuade leaders not to destroy one another, while coordinating with other AIs toward collaboration.
Kotler’s honest answer to “How would you train an AI?” is that he does not yet know. He reframes the problem from one sovereign model to millions or billions of personal AIs—each person’s lifelong Jarvis—making it a question of cooperative AI and cooperative humanity.
10. Creative brains coordinate focus and imagination instead of choosing one
Creativity is trainable, Kotler emphasizes, and declaring “I’m not creative” helps shut the process down. The core neurological contrast is between the executive-attention network, which suppresses distractions, and the default-mode network, which supports mind-wandering, imagination, self-reference, and rumination.
In most brains the two networks work in opposition: focus rises as daydreaming falls. Creative brains co-activate them, while the salience network—the hinge that selects what deserves attention—becomes unusually flexible, enabling movement between precise execution and the generation of the next unexpected idea.
Writing makes the toggle visible: a writer must focus intensely on the words in one sentence, then loosen control to discover the next line. Flow shuts down irrelevant structures while preserving the creative portion of the default-mode network.
Kotler says these differences appear in fMRI and EEG network analysis. He also says men tend to have more intrahemispheric connectivity and women more connectivity between hemispheres, while emphasizing that these are not absolutes and vary between brains.
11. Flow chemistry amplifies insight until the same dial becomes paranoia
Dopamine and norepinephrine rise near the front of flow. Both are reward and focusing chemicals; norepinephrine supplies excitement and attention, while dopamine is particularly responsive to surprise. Kotler describes using norepinephrine to close the loop in cliffhanger writing.
Pattern recognition itself produces dopamine. The small reward from solving a crossword or Sudoku problem improves signal-to-noise ratios, helping the next answers arrive in a flurry. Kotler says this is one reason flow increases creativity: dopamine and norepinephrine make additional patterns easier to notice.
The mechanism has no automatic truth filter. Turn pattern recognition too high and creativity shades into conspiracy or extreme paranoia: the brain links distant facts and rationalizes the result because it is a meaning-making machine.
Anandamide, part of the endocannabinoid system, amplifies lateral thinking. Meanwhile, flow’s efficiency exchange deactivates large portions of the prefrontal cortex, weakening the networks that maintain self and time. Past, present, and future are pushed together into what poets call “the eternal now.”
12. Live performance reveals how reduced self-monitoring unlocks synthesis
Diamandis remembers being handed a microphone without preparation, entering flow, and observing himself produce ideas he liked despite having no plan. The experience became his anchor for recognizing fluent performance without the usual internal supervision.
Public speaking contains numerous flow triggers, Kotler says, which is why accomplished speakers can finish and barely remember what they said. A truly novel question is especially valuable because it “kicks your head sideways” and forces distant data points into a new answer.
Diamandis prefers Q&A to his prepared keynote for precisely that reason. Kotler turns the episode’s own unsolved prompt—how to train an AI—into an honest “I have no idea,” followed by a commitment to think about it, perhaps through a future science-fiction novel.
13. Anxiety decides whether a team reaches for novelty or safety
Kotler’s favorite structure, the anterior cingulate cortex, helps arbitrate between pattern recognition and lateral thinking. Anxiety is the lever: as fear rises, the brain asks for the safe, logical, previously successful answer rather than the unfamiliar creative move. A scarcity mindset therefore keeps modern workers operating like threatened hunter-gatherers.
Kotler’s hiring rule is to evaluate “the person they are normally and the person they are when they’re afraid.” He favors trial periods that reveal how someone behaves while stressed, failing, and remaining responsible for creative decisions. Diamandis says he likes hiring professional action-sports athletes and people from the military because they are practiced at staying calm in extreme situations.
Kotler sees three AI responses among creatives: rejection after disappointing early trials, focused mastery of selected tools, and broad experimentation across tools. His preferred entry point is curiosity: get on the system, play with it, and ask it to teach you how to use it rather than starting with prompting.
Ten minutes daily is his compounding unit. He used short blocks to learn drawing and plans to move next to piano. AI may eventually perform either skill, but building the human capability can still improve the human-AI combination.
14. Mindsets are tunable filters on novelty, risk, and opportunity
Creativity is combinatory: new information connects with existing material to produce something novel and useful. Yet attention is distorted by a negativity bias that Kotler puts near nine negative inputs for every positive one; fear favors familiar old patterns rather than the novel material creativity needs.
A gratitude practice can reportedly shift the ratio toward five or six negative inputs per positive one, roughly doubling the novelty admitted into awareness. Biases, frames, and mindsets are therefore “tuning knobs,” not immutable defects.
Diamandis likens the brain to an LLM or neural network trained by repeated data. He recommends deliberately choosing inputs that reinforce abundance, exponential, moonshot, longevity, gratitude, curiosity, and purpose-driven mindsets.
Kotler once believed only growth-versus-fixed mindset had a real neurological basis and assumed Diamandis’s wider categories were motivational language. His correction is explicit: “I was wrong. You were totally right.” Mindsets are neurological phenomena that can be trained, and “if you don’t learn to train your brain, your brain is going to play you.”
15. Reframing converts the chemistry of fear into usable energy
Frames are momentary and conscious; repeated frames become durable mindsets and eventually operate almost like biases. Kotler’s simplest example is parking three blocks away: the same fact can be experienced as an irritation or reframed as free exercise toward 10,000 steps.
Anxiety, curiosity, excitement, and boredom sit on a norepinephrine spectrum: none produces boredom, a little produces curiosity, more becomes excitement, and too much becomes anxiety. That makes anxiety easier to redirect into excitement than to eliminate.
Kotler recounts a Harvard experiment in which saying “I’m excited” three times reduced stress more effectively than seven minutes of breathwork and meditation. He also says women over 40 in the study sometimes could not identify excitement after cultural pressure to repress it.
Diamandis extends reframing into a future Jarvis that knows a user’s emails, conversations, and history. With long-context memory, a “cognitive-bias alert” could identify recency, familiarity, confirmation, or negativity bias and surface contradictory evidence.
16. Consciousness research moved from forbidden topic to applied neuroscience
Kotler has meditated daily since 1986, yet after roughly 20 to 25 years remained convinced he was doing it wrong because the promised “fireworks” never came. Friends placed him in a 3T fMRI, where his brain looked like those of long-term meditators.
He recalls that 1990s researchers could scarcely discuss consciousness, let alone secure funding for meditation, flow, psychedelics, or mystical experience. The funding wedge was demonstrating that spirituality or religion was beneficial, after which the underlying states could be studied more openly.
The cohort included Rick Doblin, Richie Davidson, Dan Goleman, Kotler’s mentor Andrew Newberg, and Christof Koch, who publicly discussed consciousness after partnering with Francis Crick. Their approaches included psychedelics, meditation, spiritual experience, and speaking in tongues.
DMT research supplies Kotler’s reminder that set and setting matter. He describes Rick Strassman’s continuous intravenous DMT work, in which hospital participants reportedly saw gray aliens and imagined invasive experiments rather than the “machine elves” common in other DMT accounts.
17. Flow is a four-stage cycle that must be designed into the calendar
Peak-performance basics precede exotic interventions: seven to eight hours of sleep, respect for circadian rhythm, and daily nervous-system regulation. Kotler normally sleeps between 8 and 9 p.m. and wakes between 3 and 4 a.m.; he warns that modern schedules particularly punish night owls.
His regulation menu scales with available time: five minutes for gratitude, six for reading a novel, roughly seven to 11 for breathwork, and longer for walking in nature, exercise, or sauna. During COVID, Flow Research Collective employees had to complete at least three such practices daily because elevated stress consumed performance capacity.
Training then covers 28 known flow triggers—roughly 12 individual and 16 group triggers—and the four-stage cycle of struggle, release, flow, and recovery. Flow is not binary, and permanent flow is not how the brain works.
During struggle, the conscious brain loads information until frustration peaks. Kotler cites University of Michigan work suggesting greater frustration can mean a solution is closer, but only if followed by release: a low-grade physical activity such as a walk in nature, not an intense workout.
18. Small output rituals and a 100-person network compound creative work
Kotler writes from 4 a.m. until 7:38, then hikes with his dog; the walk functions as release if writing remained in struggle and recovery if flow occurred. With top performers, his goal is usually modest: identify two or three practices to start, two or three to stop, and two or three activities to move to better times.
His most durable production rule came from Johns Hopkins teacher Stephen Dixon, whom he describes as an exceptionally prolific fiction author with roughly 650 published works despite teaching and caring for a severely ill wife: “I write a page a day, and I edit what I wrote the day before.” Kotler starts books near 500 words daily, rises toward 700–800, and finishes near 1,000.
He also goes “bed to desk in under five minutes.” The brain wakes in alpha, near the alpha-theta boundary associated with flow; gamma-wave insight is coupled to theta. That routine helped produce 16 books, about five million published words, and a comparable volume of unpublished work.
The Alliance packages these ideas into an eight-month program for about 100 “super-creatives,” launching in June with three live events, six-week training cycles, AI instruction, flow science, and support for finishing an “unfinished masterpiece.” Kotler chose 100 for cross-pollination and because creativity is “a cooperative sport.”
Loneliness is the commercial and psychological wedge: “another word for world’s leading expert is nobody around to talk to.” One wealthy friend discussing the program estimated that off-days caused by loneliness cost him $5 million annually. Companies may also send groups of senior creatives as continuing education.
The Alliance is intended to carry work beyond ideation through editing, publishing, marketing, and launch. Kotler says writing a book is only half the triathlon and that publishers do almost nothing for authors; immediate feedback is a flow trigger, which is why he employs a personal editor and uses ChatGPT for critique and encouragement, though not to write his prose. The program costs around $330,000.