Adam Mosseri, head of Instagram, stated in a recent interview that frustration and anger toward AI is "going to get worse," and that this trend will extend well beyond creative industries. Most coverage treated this as a candid admission from a tech executive bracing for public backlash. I think that reading misses the more interesting claim embedded in the statement. Mosseri is not just predicting a sentiment trend. He is describing a structural condition: a growing gap between what algorithmically-mediated platforms promise and what participants can actually verify about how those promises are kept.

The Legitimacy Problem Mosseri Is Circling Around

Platform legitimacy has always rested on a particular bargain. Users contribute attention, content, and behavioral data. Platforms return reach, discovery, and audience. The terms of that bargain have never been fully transparent, but they were legible enough for participants to develop working models of how the system operated. What Mosseri is describing, whether he frames it this way or not, is the erosion of that legibility. When AI systems mediate more decisions about what content is shown, to whom, and why, the causal chain between a user's actions and their outcomes becomes genuinely opaque. The frustration he anticipates is not irrational sentiment. It is a predictable response to structural unintelligibility.

Kellogg, Valentine, and Christin (2020) documented this dynamic across multiple platform labor contexts. They found that algorithmic management systems concentrate interpretive authority with the platform while distributing consequence to workers. The result is what they call a "visibility trap": workers can observe their outcomes but cannot reliably infer the rules that produced them. Mosseri's forecast suggests this trap is now expanding from platform workers to platform users more broadly, and that AI is the mechanism of expansion.

Why the Awareness-Capability Gap Makes This Worse, Not Better

A common organizational response to public frustration with AI is increased transparency communication - explainer content, algorithmic literacy initiatives, disclosure statements. The assumption driving these responses is that if people understood AI better, they would be less frustrated. The research base does not support this assumption. Gagrain, Naab, and Grub (2024) found that increased algorithmic awareness does not reliably produce improved navigation of algorithmic environments. People can know that an algorithm exists, know roughly what it optimizes for, and still be unable to predict or influence their own outcomes within it.

This is the awareness-capability gap, and it has a specific implication for how platforms should interpret the sentiment Mosseri is forecasting. If users develop more awareness of AI mediation without developing corresponding capability to work within it, their frustration will not decrease with transparency. It will increase, because they now have enough information to know they are being structured by a system they cannot effectively engage. Sundar (2020) identified a related dynamic in AI-mediated communication contexts: machine agency attributions shift moral responsibility in ways that users find disorienting, particularly when outcomes are negative and causal attribution is unclear.

The Organizational Governance Gap Nobody Is Talking About

What makes Mosseri's statement organizationally interesting is that he is essentially predicting a legitimacy crisis that his own organization cannot resolve through conventional means. Platform governance frameworks, terms of service updates, and content moderation disclosures are all topographic interventions: they describe the terrain of rules without giving participants the structural schemas needed to navigate that terrain effectively. Rahman (2021) called this the "invisible cage" - the structural conditions through which platforms constrain participant behavior without those constraints being visible as constraints.

Mosseri's admission that negativity is "going to get worse" reads, through this lens, as an acknowledgment that the cage is becoming visible. When AI systems make more consequential decisions, and when those decisions produce outcomes that users experience as arbitrary or unfair, the structural conditions of platform governance stop being invisible. The question his statement raises but does not answer is what organizational form could plausibly restore legitimacy in this environment. Hierarchy concentrates interpretive authority further. Market exit is rarely a viable option for platforms with network effects. And network governance assumes a degree of participant parity that algorithmically-mediated environments systematically undermine.

What This Means for Platform Theory Going Forward

Mosseri is not offering a theoretical diagnosis, but his prediction maps precisely onto a theoretical gap: we do not have a well-developed account of how platforms maintain coordination legitimacy when the coordination mechanisms themselves become contested objects of public scrutiny. Classical coordination theory, as I argue in my own work on application layer communication, assumes that the rules of coordination are at minimum interpretable by participants, even if they are not negotiable. AI mediation breaks that assumption. When the rules are not interpretable, coordination does not simply become inefficient. It becomes politically unstable, which is a different kind of organizational problem requiring different kinds of organizational solutions.

Mosseri is right that the negativity is going to get worse. The more precise claim is that it will get worse until platform governance develops mechanisms that restore structural legibility, not just surface-level transparency. Those are not the same thing, and conflating them is likely how platforms will waste the next several years responding to exactly the crisis he is forecasting.

References

Gagrain, A., Naab, T., & Grub, J. (2024). Algorithmic media use and algorithm literacy. New Media & Society.

Kellogg, K. C., Valentine, M. A., & Christin, A. (2020). Algorithms at work: The new contested terrain of control. Academy of Management Annals, 14(1), 366-410.

Rahman, K. S. (2021). The invisible cage: Workers' reactivity to opaque algorithmic evaluations. Administrative Science Quarterly, 66(4), 945-988.

Sundar, S. S. (2020). Rise of machine agency: A framework for studying the psychology of human-AI interaction. Journal of Computer-Mediated Communication, 25(1), 74-88.