The Psychology of Technology: The Science We Need in the Age of AI

We founded the Psychology of Technology Institute in 2016 to advance an interdisciplinary field we call the "Psychology of Technology." A decade later, it has never mattered more.

The world is changing rapidly and dramatically. Just consider how different life was at the turn of the century. Socially, we weren’t networked via social media, nor were we text messaging each other regularly or meeting romantic partners through dating apps. In our work lives, gig jobs and remote work were the exception; Zoom meetings and Slack messaging unheard of. Leisure time wasn’t spent on phones. Education wasn’t an attention-competition between tablets and teachers. As we move into the future, more life-changing technology looms: artificial intelligence, robots, virtual and augmented reality, personalized gene therapy – and much more.

But is our powerful technology making us healthier and happier? With unprecedented technologies have come new psychological crises: fragmented attention, rising depression and anxiety, a loneliness epidemic, deep distrust and polarization. Modern technologies have negatively impacted our minds and we’re paying the price.

Society can’t afford to neglect psychological health as it develops even more powerful versions of AI. Protecting psychological health requires a dual approach: First, we need high-quality research to help us understand the complex, two-way relationship between psychology and technology use. And second, we need to leverage this research to build, test, and deploy technologies that prioritize positive psychological impact.

In what follows, we explain why research on the “psychology of technology” is a pressing need, how to make this type of research higher quality, how it can be used to improve life, and how our Institute (the Psychology of Technology Institute) is facilitating these goals.

The Greatest Threats to Our Future Are Psychological

In 1692, the village of Salem, Massachusetts tore itself apart. There was no plague or famine, no invasion – only fear. Over the course of a few months, a handful of strange symptoms in a few young girls escalated into a community-wide panic that ended with twenty people executed. The community had been undone by its own warped collective psychology.

While psychological dynamics can destroy a community, the reverse is also true: the psychology of a society can be its greatest asset. For example, social trust – the willingness to extend good faith to strangers – is one of the most powerful predictors of a nation’s prosperity. Economists have found strong supportive evidence: more than a third of cross-country differences in economic development are associated with differences in inherited trust alone. On the flip side, the corruption that flourishes where trust is absent carries a price tag to match, costing the global economy an estimated $2.6 trillion a year – about 5% of global GDP.

These examples illustrate the same point: all too often, the decisive force shaping a society’s fate is not the threats it faces from the outside but, rather, what’s happening on the inside in the form of its collective psychology.

This is easy to forget today, because our attention is fixed on seemingly external threats, such as artificial intelligence, a warming climate, or the next pandemic. Each is real and serious. Yet our ability to address threats such as these depends on human psychology. In the case of AI, the risks stem less from the technology itself than from how people and organizations choose to design, use (or reject), and govern it. Understanding the human mind, then, is ground zero if we want to successfully address our challenges.

Unfortunately, the goal of understanding and protecting human psychology has been largely neglected by tech developers, investors, policymakers, and institutional adopters. The cost of this neglect shows up as products designed to exploit users rather than support them: systems optimized for extraction, which decline in quality once they have captured their market. The pattern is common enough to have earned a name – enshittification. When psychology is treated as a growth tactic rather than a design principle, this is the predictable result.

What Is the “Psychology of Technology” and Why It Matters Now

The two-way relationship between mind and machine is the subject of a field that we call the psychology of technology, which we founded the Psychology of Technology Institute ten years ago to advance. The field sits at the intersection of human psychology and digital technology. It studies the antecedents of technology – the psychological forces that shape how products get designed, adopted, and used – as well as the consequences, meaning how those products reshape how we think, feel, and relate. It also examines a third, easily missed layer: the psychology that gets embedded into technologies themselves in the form of assumptions about what users want and how they should behave.

The field is related to established traditions such as human-computer interaction, human-centered design, and sociotechnical theory, and draws on all of them. What distinguishes it is the focus on psychology as the central unit of analysis rather than one factor among many.

Why does this research matter now in particular? Because the pace of technological change has accelerated, and the power of our technology to shape human minds has increased. The AI revolution is reshaping daily life faster than our understanding can keep up, and is built around the assumption that once the technical problems of AI are solved, a good future will follow. In contrast, we argue that the key to achieving a positive future is a healthy psychology.

Social media offers a helpful cautionary tale. Platforms were built and scaled without psychological guardrails, and we are still paying for that omission through fragmented attention, increased polarization, and poor mental health. Generative AI is now scaling faster than social media did, and we must close the gap between the speed of deployment and the depth of our understanding.

A Retrospective: Lessons Learned From The Consequences of Social Technologies In the Past Three Decades

Digital technology has transformed our social lives over the past 30 years. We’ve seen a large-scale movement away from in-person engagement toward online interaction: social conversations are increasingly text-based and asynchronous (e.g., text messaging, social media), romantic partner and friend selection is dictated through online marketplaces (e.g., dating apps), and LLMs are now changing how people write and talk as well as increasingly affecting people’s social and professional networks.

Individually, each of these social technologies has had enormous potential upside, generating excitement from the market and facilitating rapid widespread consumer adoption. Dating apps made it possible to access the profiles of thousands of potential mates, rapidly sort through options, and streamline the courtship process. Social media allowed us to organize our social networks in unprecedented scope, find niche audiences for our messages, and live our lives more publicly. More generally, the proliferation of online communication tools made it easy to bypass almost any barrier that used to exist for in-person interaction, from geographic distance to pandemics. The exciting promise of each new technology super-charged our embrace of it, inciting rapid adoption. Our social worlds are now hyper-networked, textual, and opportunity-rich.

But each of these initially exciting technologies introduced deep, concerning, and lasting psychological costs. Dating apps have commodified the very act of falling in love, taking one of the most beautiful and meaningful experiences of human life and turning it into an instrumental activity that prioritizes quantifiable metrics while largely ignoring difficult-to-quantify intangible characteristics such as chemistry, loyalty, or long-term compatibility. Social media hijacked our attention using a business model that keeps “eyeballs on the screen” and facilitated compulsive attachment to phones, demonstrably reducing attention spans and increasing anxiety and depression especially in vulnerable communities (e.g., children).

Modern social technologies have contributed to making our society more polarized (by prioritizing conflict-driven, morally outrageous, and emotionally activating content) and less connected (by moving interaction into dehumanized channels that facilitate cyberbullying, scalable harassment, and reduced social accountability). They have eroded our privacy (by extracting data unethically and behaviorally tracking us) and facilitated not just misinformation but epistemic instability (by perpetuating conspiracy theories and pseudoscience, reducing institutional trust and confusion about expertise). In essence, the very same technologies we were so quick to adopt have largely commodified and dehumanized social life, undermining our mental and societal health.

What went wrong? As another powerful new technology – generative artificial intelligence – is being widely adopted, it is critical to identify the lessons learned from recent digital technology transformations like social media. Three major mistakes are particularly critical to avoid:

1. Companies using our psychology against us. Although social media companies did not prioritize psychological health as an outcome, they did use our psychology to increase their profits. From the beginning, most big tech companies built robust behavioral science divisions, hiring psychologists and turning “UX” into a common and profitable career path. But their incentives were misaligned for human flourishing. Rather than understanding user psychology to facilitate longer-term psychological and societal health, the priority was maximizing short-term ad-driven profit by optimizing for engagement.

2. Scaling before rigorous, objective, and transparent testing. Many social media companies scaled quickly, well before there was scientific convergence regarding its psychological consequence. For example, it took about two decades of debate after its emergence for scientists to begin agreeing on the potential for psychological harms resulting from excessive social media use. But by that time, the product was ubiquitous and couldn’t be easily removed or reduced. Why did scientific agreement take so long? In part because academic researchers struggled to access high-quality, causal data. And although the tech companies themselves were collecting rich internal data for their own use, they weren’t making the data nor results publicly available. Scientists external to the companies had to collect their own data which was necessarily inferior and lagged behind each new feature launch.

3. A fractured and siloed research ecosystem. Several systemic challenges have hindered our ability to identify problems with certain uses of social technology. Academia is overly siloed, with faculty often not collaborating across departments despite having similar research questions, and also failing to work closely with non-academics (e.g., technologists). Research also moves too slowly to keep pace with technological change; the peer review process can take years, during which time a technology and how it is used may have entirely transformed. Finally, there is not enough connection between the scientists who produce the research and those who disseminate it (e.g., journalists) to the audiences who need it (e.g., policymakers). The pipeline from a research study to a policy, for instance, has major points of leakage, with the original investigators rarely consulted beyond a certain point.

The lessons learned from recent social technologies make it clear: We must change course to ensure that we realize the potential benefits while avoiding the harms of current and emerging AI systems.

The Psychology of Technology Institute: How We Understand the Role of Psychology in a Technology-Driven World

We founded the Psychology of Technology Institute to improve society’s understanding of the psychology of technology and use it to improve the human-technology relationship. Our network includes hundreds of behavioral scientists, including many social, cognitive, and developmental psychologists but also scholars from related fields: sociology, organizational behavior, marketing, philosophy, computer science, , communication, economics, and more. Every year, we gather together at a different university to discuss the latest science, debate, collaborate, and consider our role in building a better future. To achieve our mission, we facilitate new partnerships and conversations across the tech ecosystem in order to ask better questions, find answers more quickly, and spread insights more efficiently among scientists, tech designers, policymakers, and end users.

Our Institute has been instrumental in helping build this science. We’ve broken down silos by facilitating new company-academic partnerships, funded research, written papers to improve research methods, and created much-needed infrastructure for the field including in-person and online convenings, relevant scientific groups, books, and classes. We’ve worked with regulators and legislators, supported legal cases, mentored hundreds of students, conducted research, and built out new education programs.

Our advisors include brilliant researchers studying everything from the challenges of fragmented attention (e.g., Paul Leonardi, Gloria Mark, Jonathan Haidt, Larry Rosen), bias in digital systems (e.g., Sendhil Mullainathan, Lindsey Cameron, William Brady), social connection and mental health (e.g., Elizabeth Dunn, Johannes Eichstaedt, Jeff Hancock, Lyle Ungar), the role of technology in workplaces (e.g., Nancy Rothbard, Roshni Raveendhran, Tara Behrend, Hatim Rahman, Batia Wisenfeld) and marketplaces (e.g., Carey Morewedge, Stefano Puntoni), regulation and policy (e.g., Kamy Akhavan, Camille Carlton, David Evan Harris), AI and design (e.g., Don Norman, Pattie Maes, Stuart Russell, Pat Pataranutaporn), society and democracy (e.g., Eli Pariser, Alice Siu), human-computer interaction (e.g., Munumn De Choudhury, Jonathan Gratch, Min Kyung Lee), and decision making and behavior change (e.g., Don Moore, Jay Van Bavel, Angela Duckworth, Wendy Wood). We are fortunate to learn from these incredible people and benefit from their work.

Taken together, our activities aim to achieve two major goals: advancing high-quality research on the psychology of technology, and disseminating that research in ways that make a difference. We review the importance of each, and our pursuit of them, below.

1. Producing high-quality research on the psychology of technology

In order to leverage technology for human flourishing, we must have rigorous evidence regarding its human impact. We think that deploying technologies without testing is akin to releasing new drugs before they undergo a rigorous clinical trial system to identify harmful side effects. We should not release powerful new technologies, especially to children, before we know how different uses of these technologies will impact our psychology.

Producing high-quality research is challenging for any field (e.g., there are dozens of critiques of psychology research methods) but we think research on the psychology of technology is uniquely challenging to do well for a number of reasons.

One unique challenge is what we call the “moving target problem:” by the time a study is designed, executed, reviewed, and published, the technology under investigation may have changed materially. Moreover, in AI research applied to humans and organizations, the target moves not only because the model changes, but also because the ecosystem is dynamic: user norms, organizational policies, UI defaults, safety layers, pricing, and complementary tools evolve. This threatens replicability, interpretability, and cumulative knowledge-building—especially for researchers who care about longer-term impact rather than momentary snapshots. Our Institute is working on a new paper that reviews why the moving target problem arises, how it distorts inference, and how to solve it. We make four recommendations to produce more durable science—mechanism-first theorizing, forward-longitudinal designs, historical benchmarking, and rigorous reporting across different versions of models.

Another unique challenge for psychology of technology research is that it requires high external validity; it’s most useful when the technology is being tested in the way it would actually be used in the world, rather than a modified version tested in the laboratory. This means researchers need to have access to the technology itself, and ideally its back-end data, which often requires collaborations with technologists who sit outside of universities. To that end, our Institute regularly connects academics with industry partners. We have organized many successful academic-industry collaborations with big tech companies (Google, Microsoft) as well as smaller start-ups ranging from GoGuardian (ed tech) to Emotect.AI (emotion detection) and many more.

Relatedly, given that the questions we examine are so large and far-reaching, we think they are best pursued across disciplines. Interdisciplinary research is especially hard because it requires breaking the silos in universities and organizations and finding common languages and practices to bridge field-specific norms and methods. Our Institute tries to help bridge disciplinary divides in several ways. For example, our annual conferences are deliberately boundary-spanning, ensuring people not only meet but also deeply interact with fields that may feel far from them. We also regularly organize other events (e.g., our “Tech Talks” series during the pandemic) where interdisciplinary scholars can present and criticize each other’s research. We conduct and review research ourselves with big, interdisciplinary teams of scientists (for examples, see: here, here, here, and here).

In part because of the interdisciplinary nature of “psychology of technology” research, the resources aren’t as readily available as with more traditional fields of research. Thus, our Institute has helped to carve out resources for this specific type of work. For example, we’ve encouraged universities and companies to hire scholars who study the psychology of technology. Our members have written books to help define the field (e.g., Sandra Matz’s Psychology of Technology textbook; see a broader list here). We encourage graduate students to enter the field by giving out dissertation awards that help motivate their research and provide them with funding. (We’ve also disseminated larger funds to research teams.)

In everything that we do, our goal is to build the strongest possible foundation of robust, reliable, valid, and useful research on the psychology of technology.

2. Leveraging psychology of technology research to produce change

It’s not enough to just produce high-quality research, we also want to make sure it’s used in pursuit of long-term psychological health. As reviewed earlier, a major failure of social technologies over the past 30 years is that psychology was considered as the tech was developed and deployed, but in the wrong way. Users’ psychology was harnessed for short-term profit (e.g., designing for addiction) rather than used to motivate and achieve longer-term health. We have proposed – and presented to dozens of companies – the idea that technology needs to be purpose-driven instead of profit-driven. While profit is of course valuable, maximizing short-term gains while harming users and society in the process is not just short-sighted but dangerous and unethical. By prioritizing longer-term health and purpose, companies can “do well by doing good” – gaining traction in longer time horizons while not harming consumers’ mental health.

Our Institute helps to bring research directly to the people who can use it – and shape the way they use it. We have been closely working with academic partners, particularly the USC Marshall’s Neely Center for Ethical Leadership and Decision Making, MIT’s Advancing Humans With AI (AHA), and the University of California, Berkeley Haas, in order to do this. Below, we highlight three important initiatives:

The Neely Social Media Index and AI Index. In partnership with the USC Neely Center, three years ago we introduced a set of public, longitudinal, nationally representative panels of user experiences in America to improve societal understanding and decision making about social media and AI. There are two separate indices which consist of ongoing surveys following a panel of U.S. adults, measuring their positive and negative experiences with social media platforms and AI. The panel itself is impressive – more than 12,000 adults from all backgrounds in America that are tracked through the Understanding America Study. Collaborating with the UAS also gives us access to rich data for each participant (e.g., cognitive skills, personality, etc). For each wave, we release the results to the public and share the data with independent researchers. (For results, see here: https://psychoftech.substack.com/t/technology-indices.)

We have four objectives for these indices: 1) by disseminating key insights to leaders, policymakers, and regulators, we can inform and improve policy decision making; 2) by publicly posting the monthly results and comparing them across various products and platforms, we seek to keep companies accountable for the experiences that they create; 3) by sharing results with the public, we help consumers regulate their own use; and 4) by providing researchers and technologists with high-quality data they can use for their own analyses, we seek to advance the broader study of technology’s impact on society.

We’ve been gratified in the response so far to the indices. They’ve been covered by the media (e.g., Bloomberg, Politico, Wall Street Journal), are informing government accountability initiatives globally, are being used by partners seeking to reform technology company incentives, and are widely read within technology companies seeking to improve the value of their products. Many researchers are also using the data and publishing peer-reviewed articles.

Design Codes. Again in partnership with the USC Neely Center, in an effort led by Managing Director Ravi Iyer, we drafted a set of specific design changes, in part to help improve social media’s impact on society, building on evidence from both internal company product work and external studies. We solicited and incorporated feedback from a wide set of stakeholders, including technologists, academics, and civil society groups to improve the recommendations, resulting in a set of design principles that we have encouraged social media platforms to adopt. These standards are meant to avoid the trap of relying on controversial top-down definitions of what is or is not harmful content. Broadly, they enable greater explicit user control, protect children through better defaults, improve incentives for publishers, and prevent small groups of users from manipulating and harming others. They have influenced policy and inspired a similar design code for social AI that has also been influential in both tech and policy circles.

Open Benchmark of AI Impact on Humans. In partnership with MIT’s AHA group, we helped develop the first open benchmark initiative to measure AI’s impact on human well-being across physical, psychological, and societal dimensions. Most modern AI benchmarks measure what models can do: accuracy, reasoning, task completion, but say almost nothing about what AI does to the people who rely on it. For example, two models with identical capability scores can shape a user’s autonomy, mental health, and relationships in completely different ways, and the field has had no shared way to tell them apart. Our effort – which we call “ImpactBench” for short – is built to answer a different question: across realistic, multi-turn conversations, does an AI system support or undermine human flourishing?

ImpactBench currently evaluates 14 leading AI systems against 18 expert-submitted benchmarks spanning physical, psychological, and societal impact. Each construct is contributed by clinicians, educators, legal scholars, and community advocates through an open submission process, then tested through multi-turn adversarial simulation with demographically stratified personas: the way harms actually unfold in real conversations, not in isolated prompts. Every score is paired with reliability checks so users can see not just what we found, but how much to trust it. Our project website lets consumers view AI model aggregate scores down to the underlying evidence, allowing them to compare models across the three impact domains, drill into specific constructs like emotional dependence or cognitive autonomy, and read the actual multi-turn transcripts behind any verdict. To help make this tool even more accessible to end users, we’ve created “Nutrition Labels for AI.” Tell us who you are: a parent, educator, clinician, or everyday user, and ImpactBench generates a nutrition label calibrated to your context. You can see how each individual model performs in your focus area, where the model excels in human flourishing, and what some blind spots might be.

Conclusion: Our Vision For the Future

Despite the major challenges for humanity as we navigate rapid technological advancements, we remain “tech optimists.” We are living through an incredible time with unprecedented potential advancement through technology. The Psychology of Technology Institute is helping to build a technology ecosystem that centers human wellbeing and flourishing by developing and disseminating the research that is necessary to do so. If we succeed, we can look forward to a future where technology facilitates deeper and more intimate human connections, safer and more sustainable means of production, and improved global coordination to solve societal challenges. If you feel excited by this vision and our mission, we encourage you to get involved in one or more of the following ways:

• Keep updated about our work - subscribe to our newsletter, LinkedIn, and YouTube channel.

• Join the network - if you are a researcher in a relevant field. (If you join, you’ll be added to our internal email list, you’ll be listed on the website, and you’ll receive first priority for our events.)

• Engage with us - whether you want an expert to consult or give a talk or training, want to collaborate on research or dissemination, or develop another tailed engagement.

• Help fund the work of the Institute.

Together, let’s develop a healthier human-technology relationship and build a future we’re excited about.