Key Takeaways
Treat tech as a religion when it starts answering life's biggest questions
More than devices. Epstein uses religion to mean the beliefs, institutions, rituals, values, and communities through which people make meaning and bind themselves together. By that definition, technology is not merely an industry: it organizes work, relationships, identity, and expectations for the future. The book opens with Constantine, who adopted Christianity partly because he judged it a stronger ally in a struggle for power. Epstein sees a parallel in tech leaders who present their products as instruments of connection, progress, and even salvation.
A useful comparison. The claim is not that every technology is a faith or that tech has a formal church. It is that tech now supplies shared stories and daily practices on a scale few institutions can match. Its promises deserve the same scrutiny as religious promises.
What's useful about the analogy is its focus on meaning and power, not on whether believers are irrational. It asks who gets to define progress, whose lives are organized around that definition, and who benefits from the arrangement. The comparison also has limits: technology includes public infrastructure, medical tools, and ordinary craft, not just the corporate platforms Epstein chiefly criticizes. Treating all of it as one religion could blur important differences. The metaphor works best as a diagnostic question: what beliefs and authorities does a particular technology ask people to trust?
Question promises of digital salvation before they excuse present-day suffering
Heaven and catastrophe. Tech doctrine, Epstein argues, often pairs a dazzling future with a terrifying alternative. Ray Kurzweil imagines technology overcoming death; some longtermists focus on vast populations of future digital beings; AI-risk advocates warn that advanced systems could destroy humanity. Toby Ord's estimates, cited in the book, put the chance of extinction from unaligned AI at roughly one in ten over the next century. These are forecasts, not established facts, yet they can shape careers, donations, and policy priorities.
Bring the future back down to earth. Epstein questions a mindset that treats present injustice as secondary to speculative future gains. He points to tech-driven labor exploitation and proposals to build exceptional, lightly governed cities in poorer countries as reminders that visions of progress can conceal old patterns of extraction.
Long-range thinking has a serious ethical case: decisions made now can affect people who cannot yet speak for themselves. The problem is not caring about the future, but letting an uncertain forecast acquire the authority of certainty. Decision theory offers a useful discipline here: compare scenarios, track confidence, and ask what actions remain worthwhile across multiple possible futures. That approach can take AI dangers seriously without making speculative digital paradise the measure of every present-day need. Future generations matter, but so do people bearing the costs of the systems being built now.
Judge tech by who receives power, wages, safety, and a voice
Meritocracy can hide hierarchy. Tech often presents itself as open and fair, yet Epstein documents unequal access to leadership, capital, and protection. One analysis he cites estimates that white men control 93 percent of venture-capital dollars. Women and people of color have also been pushed out of technical roles as those roles became more prestigious, while women of color performed much of the painstaking early work of computer manufacturing.
Follow the costs. The book also describes biased search results and facial-recognition systems that perform worse for some demographic groups. Gig workers, content moderators, and factory workers make digital services function while receiving less security and recognition than the companies they support. A technology's benefits cannot be judged only by its usefulness to customers or its returns to investors.
This approach connects technology ethics to political economy. A system can be highly efficient for its owners while transferring risk to workers or communities with little bargaining power. That is an externality, a cost excluded from the product's price but not from real life. Representation in design matters, but it is not sufficient by itself: diverse teams can still operate under incentives that reward extraction. Strong evaluation therefore looks at decision rights, pay, recourse, and who can refuse participation, as well as whether the interface works as intended.
Interrupt phone rituals by facing the feelings they help you avoid
Checking becomes ritual. Epstein cites a survey estimating that Americans open their phones 344 times a day. He describes his own pattern of refreshing email and social feeds to escape anxiety, including attending a recovery group when Internet use began disrupting his work and relationships. The book connects compulsive checking to intermittent rewards: unpredictable likes, messages, and updates can make people return repeatedly, much like variable rewards in gambling.
Change the need, not just the device. A simpler phone helped Epstein notice how boredom and discomfort drove his checking, but a weekend at a phone-restricted camp showed him something more powerful: when meaningful human connection is available, screens can lose their pull. The practical task is to notice the feeling behind the urge and make room for a more nourishing response.
The link between avoidance and compulsive use is valuable because it shifts attention from personal weakness to emotional function. A notification can offer brief relief without resolving the anxiety that prompted the check. Still, screen time alone does not prove addiction or harm, and research on effects varies by age, activity, and circumstance. A better question than how many hours is whether use displaces sleep, attention, relationships, or chosen goals. Tools such as limits can help, but they are more likely to last when paired with social support and alternative sources of reward.
Treat surveillance as a political system, not a neutral gadget
Visibility is unequal. Epstein's Detroit account traces surveillance to a history of racial exclusion, from the destruction of Black Bottom to a police unit that used deceptive tactics and killed Black residents. In the present, Project Green Light places police-connected cameras across the city. The technology arrives in a community already shaped by displacement and distrust, so its meaning cannot be reduced to whether cameras deter some crime.
Errors have consequences. The book cites a National Institute of Standards and Technology study finding that some facial-recognition systems misidentified Black and Asian people at rates 10 to more than 100 times those for white men. It also describes surveillance used to track protestors and migrants. Data collection changes who can move, speak, and organize without being watched.
Surveillance is often defended through a simple exchange: give up some privacy for safety. That framing misses who defines safety, who is watched most, and what happens when the system is wrong. The history in Detroit shows why community consent and meaningful oversight are not optional extras. At the same time, rejecting every camera or data tool as identical can obscure legitimate uses, such as documenting harm. The sharper test is whether monitoring is limited, accountable, and proportionate, with affected communities able to challenge both its purpose and its errors.
Support tech's dissenters, even when victories are temporary
Criticism can change the terms. Labor scholar Veena Dubal challenged ride-hailing companies' classification of drivers as independent contractors and helped make labor conditions part of the tech debate. California's AB 5 briefly strengthened worker protections, but companies spent more than $200 million supporting Proposition 22, which restored contractor status for app-based drivers. The reversal showed how quickly corporate power can undo a reform.
Dissent needs company. Driver Annette Rivero organized with other workers after discovering that bonuses and app prompts could encourage longer shifts while pay fell. Epstein also describes critics facing harassment and intimidation. Their work may not win each policy fight, but it makes hidden costs visible and connects workers who might otherwise see themselves as isolated. Supporting dissent means taking their evidence seriously and sustaining coalitions beyond a single news cycle.
The episode illustrates a classic imbalance between concentrated and dispersed interests. A company can coordinate spending and messaging; each worker has limited time, money, and access to decision-makers. Organizing helps counter that asymmetry by turning individual experiences into evidence of a shared pattern. Yet dissent should not be treated as automatically correct simply because it challenges power. Durable reform depends on clear claims, transparent evidence, and alliances that include the people most affected. Criticism is not a substitute for policy design, but policy design without criticism can easily serve the status quo.
Build technology around human flourishing, not proof of human obsolescence
Worth is not output. Epstein connects tech's drive to optimize and transcend human limits with a deeper fear that people are valuable only when they achieve something exceptional. The birth of his son gave him a vivid counterexample: a child is worthy of care before accomplishing anything. That conviction shapes his idea of tech humanism, which asks how human values are encoded in machines and scaled through automation.
Expose the people behind the machine. Researcher Mary Gray's term ghost work describes human labor hidden behind supposedly automated services, including data labeling and content moderation. Social worker Desmond Patton argues that communities and ethical expertise should inform algorithm design. Together, these examples shift the question from whether a system is advanced to whether it supports people's dignity, security, and ability to care for one another.
A human-centered standard is stronger when it measures material outcomes rather than appealing language. Does a system reduce danger, expand agency, and share its benefits, or merely describe itself as responsible? This connects to human-computer interaction, where usability is understood as more than ease of clicking: the design must fit people's actual contexts and needs. The framework should also avoid romanticizing humanity. People and institutions can cause harm, too. The point is not to preserve every task for humans, but to ensure that automation serves human purposes and does not make workers invisible or disposable.
Find a reform community, then hold its leaders accountable
Belonging enables action. All Tech Is Human, the organization Epstein profiles, creates events, mentoring, career resources, and working groups for people seeking a more responsible technology sector. By early 2024, its Slack community had about 7,000 members across 88 countries. Participants described gaining professional connections as well as a sense that others shared their concerns.
Community is not a free pass. Epstein also questions the group's choices to host corporate speakers and accept major grants. At one event, a Meta representative faced little challenge over the company's record, while a former employee publicly pressed the need to confront the business model itself. A community can help people resist isolation, but its credibility depends on whether it applies its own standards to its funders, partners, and leaders.
Communities can convert private unease into collective capacity. They offer the trust and repetition that one-off campaigns lack, and they can help newcomers find work that aligns with their values. But organizations built around reform face a familiar governance problem: the resources that allow them to grow may come from institutions they hope to change. Transparency about funding, room for internal disagreement, and independent decision-making are practical safeguards. Belonging should make scrutiny easier, not harder. The strongest community is not one without conflict, but one that can address conflict without silencing its members.
Stay agnostic: use technology without surrendering judgment to its prophets
Not knowing can be a stance. Epstein's tech agnosticism is neither automatic rejection nor uncritical faith. It means recognizing uncertainty while judging technologies by their consequences for people. The book links this stance to Lesley Hazleton's account of agnosticism as an active engagement with what cannot yet be settled, rather than indecision.
Skepticism needs standards. Researchers David Rand and Gordon Pennycook found that overconfidence is associated with belief in conspiracy theories, and that prompting people to consider accuracy can reduce the sharing of misinformation. Epstein's conclusion is that doubt should apply to tech evangelists and to skeptics themselves. Keep asking what evidence supports a claim, who benefits if it is believed, and what evidence would change the judgment.
This is a useful account of epistemic humility: the capacity to hold a belief provisionally and revise it when evidence changes. It differs from blanket doubt, which can make well-supported findings seem equivalent to baseless claims. Scientific reasoning depends not on refusing conclusions, but on matching confidence to evidence and remaining open to correction. That discipline is especially important when charismatic leaders offer certainty about distant futures. Agnosticism becomes actionable when paired with standards for evidence, a willingness to update, and the courage to make decisions before certainty arrives.
Analysis
The book's central move is to treat technology as a meaning-making power, not merely a collection of tools or an economic sector. Its religious analogy gathers otherwise separate phenomena into one view: salvation narratives, elite authority, attention rituals, surveillance, and the hope of transcendence. The structure matters. It begins with belief, turns to everyday practice and material hierarchy, then looks for reform through dissent, humanism, and community. The author's memoir-like scenes keep this large argument grounded in embodied experience: anxiety in front of a phone, unequal access to basic facilities, and the intimate value of care.
The analogy is most persuasive when it reveals how moral language can legitimize power. A company can promise connection while depending on surveillance, or celebrate innovation while concealing labor. The comparison also raises a useful standard: assess institutions not only by their stated ideals, but by the habits they cultivate and the people who bear their costs. Its limit is breadth. Technology is not a unified faith, and its uses range from lifesaving medical tools to addictive platforms. The book recognizes this, but its recurring religious frame can still make unlike systems seem more coherent than they are.
The proposed alternative, tech agnosticism, is not a specific policy program. It is a posture of disciplined uncertainty joined to moral commitment. That combination matters because uncertainty alone can become paralysis, while conviction alone can become another form of evangelism. The book's most actionable implication is therefore institutional as well as personal: cultivate communities where people can question powerful claims, include those affected by design decisions, and judge success by human consequences. Reform requires both skepticism toward promises and constructive work on what should replace harmful arrangements. The argument does not resolve every tradeoff, but it makes the right ones harder to ignore.
People Also Read
Glossary
Tech religion
Tech as meaning systemEpstein's term for technology's broad role in organizing beliefs, values, identities, rituals, institutions, and hopes for the future. The comparison does not mean that technology has a formal church or that every tool is religious. It highlights how tech shapes daily life and offers stories about progress, salvation, danger, and human purpose.
Tech agnosticism
Critical openness to technologyA stance that avoids both automatic faith in technological progress and blanket rejection of technology. It accepts uncertainty about whether particular tools or systems will help, then evaluates them by their effects on human dignity, justice, agency, and shared well-being. The stance requires evidence and judgment rather than deference to tech leaders or fashionable skepticism.
Tech humanism
Technology guided by human valuesAn approach that treats technology as a human creation whose design and deployment reflect human choices. It asks what values are built into systems, what effects are scaled through automation, and whether technology supports human flourishing. In the book, tech humanism includes work by researchers, organizers, and practitioners who seek more just and humane technologies.
Tech theology
Tech's account of life's meaningThe beliefs and assumptions through which tech culture explains human purpose, progress, and the future. Epstein uses theology broadly, not only as talk about gods, but as reflection on what matters most. Tech theology includes confidence in innovation as a route to control, transformation, and a better future.
Digital Puritanism
Merit framed as moral worthEpstein's comparison between tech culture and Puritan ideas that treated success as evidence of being among the chosen. In contemporary tech, exceptional achievement and entrepreneurial creativity can function as signs of worth, while those who struggle are treated as less deserving. The concept connects meritocratic language to inherited privilege and social exclusion.
Technochauvinism
Technology treated as the solutionA term used by researcher Meredith Broussard for the belief that technology is always the best answer to social problems. In Epstein's account, this outlook often accompanies faith in meritocracy, claims that computer systems are objective, and the expectation that wider use of technology will produce a more equitable society without addressing underlying power relations.
Ghost work
Hidden human labor behind automationA term used by Mary Gray and Siddharth Suri for work that enables digital services but is often obscured from users. It includes tasks such as labeling data, moderating content, and processing information that companies may present as automated. The term makes visible the people whose labor allows systems marketed as artificial intelligence to function.
Experiential avoidance
Escaping difficult inner statesA psychological pattern in which a person tries to avoid uncomfortable thoughts, emotions, or bodily feelings, even when the avoidance causes longer-term harm. Epstein applies the concept to compulsive technology use: checking a device can briefly relieve anxiety or boredom, reinforcing the behavior without addressing the underlying distress.
Collective effervescence
Energy created through shared ritualSociologist Emile Durkheim's term for the heightened feeling of connection and shared significance that can arise when people participate in common rituals. Epstein uses it to explain why repeated communal practices can bind people together more powerfully than agreement on doctrine alone, and why technology's shared habits can shape culture.
Augmented Undercommons
Liberatory community using technologyPayton Croskey's proposal for a parallel community where people refuse to accept surveillance as the price of participation and use technology in ways that support freedom and collective flourishing. It adapts the idea of the undercommons, a community formed in relation to but not fully absorbed by elite institutions, to digital life and technology design.
Luxury surveillance
Surveillance sold as convenienceChris Gilliard's critical phrase for technologies that make extensive monitoring appear desirable, convenient, or prestigious. Examples include networked cameras and data systems used in homes, public spaces, and policing. The term emphasizes that surveillance can be packaged as a consumer benefit while increasing the exposure and vulnerability of people with less power.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.