This Insight Brief is brought to you by Infegy’s partnership with Joseph Bayer’s students at The Ohio State University. We work with these students to dig into a particular topic of their interest and encourage deeper insights.
When AI tools first entered everyday conversations, they were often framed as accessible solutions for support, productivity, and guidance. In the context of mental health, this early framing positioned AI as a potential companion, offering users a way to process emotions, seek advice, and navigate personal challenges.
However, the conversation around AI and mental health reflects a more complex reality. Engagement does not grow steadily over time but instead spikes during moments of heightened collective anxiety. At the same time, the overall tone of these conversations remains predominantly negative.
Rather than focusing on AI as a reliable support tool, many discussions are shaped by broader uncertainty, where themes of crisis, instability, and vulnerability appear alongside mentions of help and therapy.
Using Infegy Starscape, we’ll examine how AI is being positioned within mental health conversations today. The findings suggest that AI is not yet understood as a clearly defined solution but instead exists within an ongoing tension between support and uncertainty.
Figure 1: Social universe (total posts to social media) around AI and mental health conversations (January 26, 2026 – April 27, 2026); Infegy Social Dataset.
As is typical in these insight briefs, we begin with a look at post volume in the social universe, or how often people post about a topic, as a proxy for how much collective attention it holds.
Figure 1 shows the social universe around AI and mental health holding at a high, steady level throughout the window, roughly a quarter-million posts per period, rather than building gradually. We see an early rise to about 340,000 posts in February, a long plateau through late winter, and then the conversation's largest peak, near 378,000 posts, in late March. (The sharp drop at the far right reflects an incomplete final day of data, not a real decline.) Thus, we can conclude that baseline conversation about AI and mental health is large and durable.
What spikes and fades is engagement, the likes, shares, and replies shown in Figure 2, which surges to its peak in early February before falling back to a low, steady band. In other words, the topic is always being discussed; what changes is how viral and emotionally charged that discussion becomes, as expressed through users’ liking and sharing such posts.
A closer examination of posts during this period, as in Figure 3, reveals that the spike is driven less by discussions of AI as a mental health support tool and more by viral, emotionally charged content centered on broader uncertainty. Frequently used terms such as “crash,” “recession,” “WW3,” and “industrial revolution” indicate that AI is being discussed within a wider context of societal and economic anxiety. This pattern suggests that AI plays a dual role in mental health conversations, not only as a potential source of support, but also as a focal point for collective stress and uncertainty. Rather than reflecting increased adoption of AI for emotional support, spikes in engagement are often driven by a collective anxiety of how AI is shaping the future of work and life as we know it.
Figure 3: Dominant terms during the early February spike in AI and mental health conversations (January 26, 2026 – April 27, 2026); Infegy Social Dataset.
The narrative web shows that conversations about AI and mental health are not organized around simple optimism or rejection. Instead, the discussion centers on a number of themes that describe AI as a possible source of support and the emotional and clinical concerns that people may attempt to treat using it.
Support-oriented clusters include terms such as “therapist,” “helpful,” “better,” “trust,” “chat,” and “GPT,” suggesting that some users discuss AI as a tool for guidance, emotional processing, or access to support. However, additional clusters appear around more negative terms and themes including “addiction,” “crisis,” “lonely,” “alone,” “problem,” and “fear.” This reveals the depth of topics that people turn to AI seeking to treat, which, in turn, are often terms that are cited in conversations around the dependency, reliability, and emotional safety of using AI as therapist.
Figure 4: Narrative clusters in AI and mental health conversations, with key topics colored by sentiment (January 26, 2026 – April 27, 2026); Infegy Social Dataset.
Rather than showing that AI is replacing mental health care, the narrative structure suggests that people are still negotiating what concerns AI could play a role in treating. AI appears in the conversation as both a low-barrier support option and a wide-reaching therapeutic tool whose limits remain unclear.
Net sentiment trends show that conversations around AI and mental health are predominantly negative, with most data points falling below the neutral baseline. While sentiment fluctuates over time, positive spikes are brief and do not persist. Several sharper dips indicate moments of intensified negative reactions, while positive peaks are less sustained. Overall, the pattern suggests that negative sentiment remains the dominant tone despite ongoing variability.
Negative sentiment within AI-related mental health conversations is less about outright rejection of the technology and more about the emotional context in which it is discussed. Frequently occurring terms such as “help,” “issues,” and “problem” suggest that many posts are framed around personal struggles and requests for support. At the same time, the presence of terms such as “therapist,” “human,” and “model” indicates that users are actively negotiating AI’s role relative to traditional forms of care. This pattern suggests that negative sentiment is driven by uncertainty and vulnerability, rather than simple opposition to AI itself.
Figure 6: Dominant negative themes in AI and mental health conversations (January 26, 2026 – April 27, 2026); Infegy Social Dataset.
Content categorization suggests that AI-related mental health conversations are not isolated within health-focused discourse but are instead embedded within broader technology and computing discussions. While mental health appears as a recurring theme, it represents only a subset of a larger conversation dominated by AI functionality, tools, and technological development. This pattern indicates that discussions of mental health are often framed through a technological lens, where emotional concerns are intertwined with conversations about AI capabilities and use cases.
Figure 7: Content subject distribution of AI-related mental health conversations (January 26, 2026 – April 27, 2026); Infegy Social Dataset.
The data suggests that AI is becoming a consistent point of reference in how people discuss mental health, but not as a clearly defined solution. Instead, AI appears within conversations shaped by uncertainty, where it is positioned both as a potential source of support and as part of broader concerns about instability. As engagement trends show, attention to AI in mental health contexts is not steady but concentrated around moments of heightened anxiety. During these periods, AI is discussed less as a standalone tool for emotional support and more as part of wider narratives involving economic stress, societal uncertainty, and personal vulnerability.
Taken together, these patterns indicate that AI’s role in mental health conversations remains unresolved. Rather than replacing traditional forms of care, AI is being integrated into how individuals articulate and process uncertainty, suggesting that its function is still being actively negotiated within everyday discourse.
1. Engagement is driven by anxiety, not adoption.
Spikes in conversation are not linked to increased use of AI for mental health support, but to moments of heightened uncertainty. During these periods, AI is discussed within broader narratives of economic and societal instability, rather than as a standalone solution.
2. Conversations are defined by tension, not consensus.
AI is framed simultaneously as a source of support and a source of concern. Discussions around “help,” “therapist,” and “chat” coexist with language related to “crisis,” “lonely,” and “problem,” indicating that users are actively negotiating AI’s role in emotionally sensitive contexts.
3. Negative sentiment reflects vulnerability, not rejection.
While sentiment remains predominantly negative, this tone is largely driven by personal struggles and uncertainty rather than opposition to AI itself. Conversations frequently center on requests for help and unresolved concerns, positioning AI within a broader context of emotional need rather than technological resistance.
If you are interested in the types of insights provided by this Insight by Infegy, please schedule a demo.