Why not try a little helpscrolling for a change?

Q&A with Dr. Nitin Agarwal

Editor’s note: A federal judge recently found Meta and YouTube liable for negligence, ruling that their social media platforms were designed to be addictive and harmed a young user’s health. For more on what this means—both for tech companies and their social media customers—we put a few questions to frequent contributor Dr. Nitin Agarwal, whose COSMOS Research Center at UA Little Rock researches social media for the likes of NATO, U.S. National Science Foundation, and for specific units of the U.S. Army, U.S. Navy, and U.S. Air Force. In fact, Dr. Agarwal fielded these questions via email as he traveled to presentations and meetings from the Netherlands to India to Austria. We think you’ll find his comments both compelling and sobering.

_______________________________

First, from what you know about social media, please explain how these algorithms caused users to become addicted.

At the COSMOS Research Center, our research shows that social media algorithms are optimized for engagement, not well-being. A.I.-driven recommender systems learn user preferences and exploit psychological triggers like variable rewards, social validation, and novelty. Features such as infinite scroll and autoplay create continuous feedback loops that reduce stopping cues. Supported by projects funded by the U.S. Department of War, we’ve observed how these systems scale behavioral reinforcement globally, effectively nudging users toward compulsive use patterns.

What do you think this “addiction” feels like and looks like?

Social media “addiction” often manifests as a loss of control—users intend to spend minutes but remain engaged for hours. It creates a persistent pull to check notifications, seek social validation, or avoid missing out—commonly referred to as FOMO (fear of missing out). Behaviorally, this appears as compulsive scrolling, disrupted sleep patterns, and an increasing emotional dependence on feedback loops.

In our cognitive security research, we observe clear parallels with habit-forming systems, where users experience anxiety when disconnected and relief upon re-engagement. This cycle is subtle yet powerful, particularly among adolescents whose cognitive control systems are still developing. Importantly, such behavioral vulnerabilities can be exploited by adversarial information campaigns, where malicious actors strategically trigger emotional and cognitive responses to influence target populations at scale.

Can you discuss a couple of examples of malicious actors’ taking advantage of these “behavioral vulnerabilities” for their own purposes?

Our studies show that adversarial actors routinely exploit platform-driven behavioral vulnerabilities. In the Indo-Pacific region, for instance, we observed anti-U.S. coordinated influence campaigns amplify emotionally charged or polarizing content to trigger outrage and prolong engagement, thereby increasing their algorithmic reach. In crisis situations, such as natural disasters and pandemics, malicious actors exploit fear and uncertainty to accelerate the spread of misleading narratives.

Additionally, we observe algorithmic manipulation by adversaries, who intentionally game A.I.-based content curation systems through coordinated posting, bot amplification, and engagement farming to steer algorithms toward disproportionately promoting specific narratives. In effect, they exploit both human psychology and the underlying A.I. systems, turning platform design into a mechanism for large-scale cognitive influence.

We live in an increasingly contentious world, and some have pointed to the Internet as being largely to blame. Do you agree with that? If so, knowing what you now know about the engineering of social media algorithms, how much of the blame should tech companies bear for our fractured society?

The Internet, and social media in particular, has amplified societal divisions, but it is not the sole cause. These platforms accelerate and amplify existing tensions by prioritizing content that drives engagement, often including polarizing or emotionally provocative material—prioritizing virality over veracity. Companies like Meta and YouTube are not responsible for creating societal fault lines, but their algorithmic designs can intensify them. From a research standpoint, responsibility is shared, but platforms bear significant accountability for engineering systems that systematically amplify division at scale without sufficient safeguards. Bottom line: Technology like A.I. doesn’t define us, it reflects us.

What would you say this case reveals about the true purpose of social media, from the tech world’s point of view?

This case highlights that the dominant business model of platforms like Meta and YouTube is rooted in attention maximization. From a technical standpoint, A.I. systems are designed to optimize time-on-platform because attention directly translates into advertising revenue. The verdict underscores that user engagement is not incidental—it is engineered. While platforms offer connectivity and expression, their underlying incentive structures prioritize growth and retention, often without sufficient safeguards for long-term user well-being.

Your research is on how social media is applied en masse, socially and politically, right? How does this judicial verdict reflect on the way you’ll now view future social media campaigns of various kinds?

This verdict reinforces a key insight from our work in social computing and cognitive security: Platform design shapes collective behavior at scale. Whether in public health messaging or socio-political campaigns, algorithms can amplify influence in unintended ways. Moving forward, I will place greater emphasis on how engagement-driven systems may distort information exposure and emotional responses. Our projects in international contexts show that these dynamics can affect societies, polarization, and crisis response, making it critical to evaluate not just the content but also the algorithmic infrastructure that delivers it.

In light of this ruling, we’re hearing comparisons with the way lawsuits took down Big Tobacco. How would you imagine such a crackdown on social media happening today? Will the major tech companies be forced to change their addictive algorithms?

A similar trajectory is possible, though more complex. Like tobacco, internal knowledge and design intent are now under scrutiny. We may see regulatory pressure around algorithmic transparency, age protections, and duty-of-care standards. Leaders like Arkansas Governor Sarah Huckabee Sanders have already taken steps to address youth exposure and online harms, signaling growing policy momentum. As a member of the Governor’s A.I. Taskforce, I see increasing interest in responsible A.I. governance. While change may be gradual, sustained legislative and litigative pressure could compel platforms to redesign high-risk features.

How do you see that changing social media as you’ve come to know it?

If reforms take hold, social media may evolve from frictionless, high-engagement environments into more deliberate, user-centric ecosystems. We could see constraints on features like autoplay, greater algorithmic transparency, and stronger safeguards for minors. A.I. systems may be reoriented toward promoting “healthy engagement” rather than maximizing time-on-platform.

Our research at COSMOS suggests that such shifts would likely reduce virality and compulsive use while enhancing trust and long-term platform sustainability. Ultimately, this transformation could move social media beyond the attention economy toward a more accountable, human-centered digital infrastructure—one that is also more resilient against malicious actors seeking to exploit platform vulnerabilities for cognitive warfare.
______________________________
Nitin Agarwal, the founding director of the COSMOS Research Center at UA Little Rock, is the Maulden-Entergy Chair and Donaghey Distinguished Professor of Information Science at UALR. At COSMOS, he leads U.S. Department of War and National Science Foundation-supported projects with over $30 million in direct funding. He is a senior member and distinguished visitor of IEEE, member of ACM, fellow of Arkansas Research Alliance, fellow of Arkansas Academy of Computing, fellow of IARIA, and faculty fellow at ICSI, University of California, Berkeley. Visit https://agarwalnitin.com for more details.