HUMAN CLARITY INSTITUTE · FULL RESEARCH REPORT
The Loneliness Trade-Off
AI Companions and the Rise of Synthetic Intimacy
Abstract
Conversational AI systems are increasingly present in everyday digital life, raising questions about how people experience emotional interaction with artificial systems. This report examines patterns of AI companionship, emotional disclosure, and perceived meaning in AI interaction using data from the Human Clarity Institute’s AI Companionship & Human Connection 2025 dataset.
The findings suggest that while AI companionship remains uncommon, conversational AI may serve as a private space for reflection or expression for a minority of users. In this sample, 14% of respondents describe AI interaction as companion-like, and 12% report receiving emotional support from AI tools. At the same time, 85% report that AI interaction feels less meaningful than human interaction, indicating that human relationships remain the primary source of connection.
A notable proportion of respondents also report personal disclosure in AI interaction. 39% report high comfort sharing personal thoughts with AI, while perceptions of emotional safety are divided between those who feel more comfortable engaging with AI and those who feel less comfortable doing so.
Taken together, these patterns suggest the emergence of synthetic intimacy — emotionally expressive interaction with conversational AI systems that can feel personal without involving a human relationship. The findings also indicate a developing loneliness trade-off, where AI interaction may provide reflective or emotional relief for some users even while human relationships continue to anchor meaningful connection.
Executive Summary
This report examines how people experience emotional interaction with conversational AI systems. While public discussion often assumes that AI companions are rapidly replacing human relationships, the evidence in this dataset suggests a more complex reality.
Most respondents do not experience AI as a companion or emotional support system. Instead, conversational AI appears to function primarily as a tool or reflective interface. However, a notable minority report engaging with AI in ways that involve personal expression or emotional reflection.
These interactions do not appear to displace human relationships for most users. Respondents overwhelmingly report that human interaction remains more meaningful than AI interaction, and most say that conversations with AI rarely replace talking to other people.
At the same time, conversational AI introduces a new type of interaction that did not previously exist in everyday digital life. For some users, AI systems appear to function as a private space for reflection, idea exploration, or emotional processing.
This report describes these emerging patterns using the concept of synthetic intimacy: emotionally expressive interaction with AI systems that can feel personal without involving a human relationship.
The findings also point to what may be described as a loneliness trade-off. Conversational AI may provide moments of reflection or emotional relief for some users even while human relationships remain the primary source of meaningful connection.
The results presented here provide an early benchmark for understanding how people experience AI companionship today. As conversational AI systems evolve and public discussion around AI relationships expands, tracking these patterns over time will be important for understanding how digital technologies shape the human experience.
By The Numbers
14%
describe AI interaction as companion-like
12%
report receiving emotional support from AI tools
39%
report high comfort sharing personal thoughts with AI
85%
say AI interaction feels less meaningful than human interaction
41%
feel less emotionally safe with AI than with humans
41%
feel more emotionally safe with AI than with humans
69%
say AI rarely or never replaces talking to people
30%
report frequent loneliness
31%
report at least occasional replacement of conversations with people
Younger respondents
show higher comfort sharing personal thoughts with AI than older groups
Frequent AI users
show higher levels of disclosure and emotional interaction with AI systems
Gender differences
appear limited in this dataset relative to age and usage intensity
Introduction
Conversational AI systems have rapidly evolved from task-oriented digital tools into interactive environments capable of sustained dialogue. Systems such as chatbots, language models, and AI assistants can now respond in ways that appear conversational, reflective, and emotionally responsive.
This technological shift has introduced a new question into everyday digital life: how do people experience emotional interaction with artificial systems?
For some users, AI interaction remains purely instrumental — a way to search for information, complete tasks, or generate ideas. For others, conversational AI may become a place to express thoughts, ask questions, or explore ideas in ways that feel more personal.
Public discussion around AI companionship often focuses on extreme cases: artificial partners, AI friendships, or concerns about people replacing human relationships with machines. However, empirical evidence on how common these experiences actually are remains limited.
The Human Clarity Institute’s AI Companionship & Human Connection 2025 dataset explores this topic by examining how people perceive AI interaction across several dimensions, including:
- perceived companionship
- emotional support
- comfort sharing personal thoughts
- perceived meaning compared with human interaction
- whether AI interaction replaces conversations with people.
The results suggest a more nuanced picture than commonly portrayed in public discourse. While most people do not experience AI as a companion or emotional support system, a meaningful minority report interacting with AI in ways that feel personal or reflective.
This report introduces the concept of synthetic intimacy to describe these interactions. Synthetic intimacy refers to emotionally expressive interactions with AI systems that can feel personal or reflective without involving a human relationship.
The findings also point to what may be described as a loneliness trade-off. Conversational AI may offer a form of emotional relief or reflection for some individuals, even while human relationships remain the primary source of meaning and connection in everyday life.
What Is Synthetic Intimacy?
Synthetic intimacy describes interactions with AI systems that involve elements of emotional expression, personal reflection, or perceived understanding despite the absence of a human relationship.
These interactions differ from traditional uses of digital tools. Rather than simply retrieving information or completing tasks, users may engage conversational AI in ways that involve:
- discussing personal concerns
- expressing feelings or frustrations
- exploring ideas or internal questions
- reflecting on decisions or experiences.
In such cases, AI systems function less like tools and more like interactive spaces for reflection.
Importantly, synthetic intimacy does not imply that users believe AI systems possess emotions or consciousness. Instead, it reflects how conversational interaction can create an environment that feels personal, even when users understand that the system itself is artificial.
The data examined in this report suggest that while synthetic intimacy exists, it remains a minority experience among users of AI systems.
Do People Actually Experience AI as a Companion?
One of the central questions surrounding conversational AI is whether people experience AI systems as companions.
In this dataset, 14% of respondents rate AI interaction as companion-like, based on scores of 5–7 on a seven-point scale measuring perceived companionship. In contrast, 83% rate companionship as low, suggesting that most people do not experience AI as a substitute companion.
Similarly, emotional support from AI appears limited. 12% report receiving meaningful emotional support or comfort from AI systems, while 80% report low levels of emotional support.
These results suggest that AI companionship remains uncommon. For most users, AI interaction does not replace traditional forms of companionship or emotional support.
However, the presence of a minority who do report these experiences indicates that conversational AI is beginning to occupy a new role in the emotional landscape of digital life.
If AI Is Not a Companion, Why Do People Open Up to It?
While relatively few people describe AI as companion-like, a larger proportion report feeling comfortable sharing personal thoughts with AI systems.
In this dataset, 39% report high comfort sharing personal thoughts or feelings with AI, while 48% report low comfort doing so.
This finding suggests that AI interaction may function less as companionship and more as a private reflection space.
Conversational AI systems offer several characteristics that may make them appealing for personal reflection:
- they are available at any time
- they respond without visible judgement
- they allow users to explore thoughts without social consequences.
These features may create an environment where some individuals feel comfortable expressing thoughts that they might hesitate to share elsewhere.
The data also reveal a striking division in how emotionally safe AI interaction feels. When asked to compare emotional safety with AI versus humans, 41% report feeling less emotionally safe with AI, while 41% report feeling more emotionally safe.
This polarisation suggests that emotional interaction with AI is not experienced uniformly. For some individuals, conversational AI may feel like a neutral environment for reflection. For others, concerns about privacy, authenticity, or trust may make emotional interaction with AI feel less comfortable.
Is Talking to AI Replacing Human Interaction?
A central concern surrounding AI companionship is whether conversational AI may replace human interaction.
In this dataset, most respondents report that AI interaction does not replace conversations with people. 69% say AI rarely or never replaces talking to others.
However, a minority report some level of substitution. 31% say AI interaction replaces talking to people at least occasionally, including 5% who report very frequent replacement.
These findings suggest that while AI interaction does not widely replace human relationships, substitution behaviour does exist at the margins.
This pattern may represent an early signal of how conversational AI can begin to intersect with everyday social behaviour. For some users, AI interaction may supplement or occasionally replace conversations that might otherwise occur with friends, family, or colleagues.
At present, however, this behaviour appears limited to a minority of users.
Why Does Human Connection Still Feel More Meaningful?
Despite the emergence of synthetic intimacy, human relationships remain central to how people experience meaning and connection.
When respondents were asked to compare the meaningfulness of AI interaction with human interaction, 85% report that AI interaction feels less meaningful than human interaction.
Only 7% report that AI interaction feels more meaningful than human interaction.
This large gap suggests that while AI interaction can feel useful or expressive, it does not replicate the depth of meaning associated with human relationships.
Who Is Most Likely to Form Personal Connections With AI?
Patterns in the data also suggest that experiences of synthetic intimacy vary across different groups of users.
Younger respondents show higher comfort sharing personal thoughts with AI systems compared with older groups. This may reflect broader generational familiarity with digital communication environments, where conversational interaction with technology is already common.
Frequency of AI use also appears relevant. Respondents who report using AI tools more frequently show higher levels of personal disclosure and emotional interaction with AI systems.
In contrast, differences between men and women appear limited in this dataset, suggesting that emotional interaction with AI may not follow traditional gender patterns.
These patterns indicate that experiences of synthetic intimacy may depend more on how frequently individuals interact with AI systems than on demographic characteristics alone.
The findings suggest a loneliness trade-off: conversational AI may provide reflective relief for some users even as human relationships remain central to meaningful connection.
What Does the Loneliness Trade-Off Look Like?
The findings in this dataset suggest what may be described as a loneliness trade-off.
Conversational AI systems may provide a form of emotional relief or reflection for some users while human relationships remain the central source of meaningful connection.
Several patterns in the data illustrate this tension.
First, a meaningful minority of respondents report experiences associated with synthetic intimacy:
- 14% describe AI interaction as companion-like
- 12% report emotional support from AI
- 39% report comfort sharing personal thoughts with AI
At the same time, human relationships remain dominant:
- 85% say AI interaction is less meaningful than human interaction
- 69% report that AI rarely or never replaces talking to people
The dataset also shows that 30% of respondents report frequent loneliness, suggesting that emotional needs remain present even among populations with access to AI tools.
Together, these patterns suggest that conversational AI may offer a new space for reflection or expression without replacing human relationships. In some cases, AI interaction may provide temporary emotional relief or support, particularly when human interaction is unavailable.
However, the data indicate that most people still view human relationships as the primary source of meaningful connection.
Conclusion
Conversational AI has introduced a new form of interaction into everyday digital life. While these systems were initially designed as tools for information and productivity, many users now engage with them in ways that involve reflection, expression, and conversational exchange.
The findings in this report suggest that AI companionship remains uncommon, but that forms of synthetic intimacy — emotionally expressive interaction with AI systems that can feel personal without involving a human relationship — are beginning to appear among a minority of users.
For some individuals, conversational AI appears to function as a space for private reflection or expression. Respondents report sharing thoughts, exploring ideas, or engaging in dialogue with AI systems even while recognising that these systems are artificial. At the same time, the data show that human relationships remain the primary source of meaning and connection. Most respondents report that AI interaction feels less meaningful than human interaction and rarely replaces conversations with other people.
Taken together, these patterns suggest what may be described as a loneliness trade-off. Conversational AI may offer a form of emotional relief or reflection for some users, even while human relationships continue to anchor meaningful connection in everyday life.
Importantly, the results presented here describe patterns observed in the present dataset, not long-term outcomes. The behaviours associated with synthetic intimacy appear limited to a minority of users, and the data show strong evidence that human connection remains central to how people experience meaning and companionship.
This dataset therefore provides an early benchmark for understanding how people experience AI companionship today. As conversational AI systems continue to evolve and public discussion around AI relationships expands, patterns of emotional interaction with AI may also change.
The Human Clarity Institute will continue monitoring this topic through future research and longitudinal data collection. Comparing future datasets with the patterns observed here may help clarify whether synthetic intimacy remains a minority experience or becomes a more common feature of digital life.
Understanding how people interact emotionally with emerging technologies will remain an important part of documenting the human experience in the AI era.
AI Companionship & Human Connection Data 2025
View the summary page with verified headline statistics, key patterns, and interpretive context from the AI Companionship & Human Connection 2025 survey.
AI Companionship & Human Connection 2025 Dataset
Access the full open dataset, documentation, variable definitions, and supporting materials used to produce this report and related HCI analysis.
Data & Methods Note
This report draws on data from the Human Clarity Institute’s AI Companionship & Human Connection Survey conducted in December 2025. The study used an anonymised, self-report survey design to capture how participants described their experiences of emotional interaction with conversational AI systems and how those interactions relate to human connection.
Participants were recruited through an online research panel and provided informed consent prior to participation. Responses were collected anonymously, with no personally identifiable information retained. Findings are reported at a population level only.
The survey collected responses from 501 adults residing in six English-speaking countries: the United States, the United Kingdom, Canada, Australia, Ireland, and New Zealand. Participants were required to be fluent in English. Results reflect observations from a single survey wave and should be interpreted as indicative rather than representative.
The underlying dataset used in this report is published by the Human Clarity Institute as the AI Companionship & Human Connection 2025 dataset and is available in the Institute’s open data library, where the full dataset, variable definitions, and supporting documentation can be accessed.
Non-Diagnosis & Interpretation Boundaries
All findings presented in this report are descriptive in nature. The report does not diagnose individuals, classify behaviours as conditions, determine causes or mechanisms, or evaluate the effectiveness of any approach. It does not provide advice or recommendations.
The observations described here reflect population-level patterns derived from a single survey wave. Individual experiences may differ, and interpretations beyond what can be directly supported by the data rest with the reader.
How to Cite & Where to Go Deeper
This report is published by the Human Clarity Institute as an independent research report documenting descriptive patterns observed in a large-scale survey examining how people experience emotional interaction with conversational AI systems.
The report is intended to be cited as institute research or grey literature. It provides population-level observations and interpretive framing designed to support understanding, exploration, and context-setting across research, policy, technology, and design discussions. It does not present causal findings, predictive models, or policy recommendations.
Instead, its purpose is to document emerging behavioural patterns in human interaction with conversational AI and to introduce conceptual language — such as synthetic intimacy and the loneliness trade-off — that may help researchers and practitioners better understand how digital technologies are reshaping experiences of reflection, expression, and connection.
For academic or analytical work requiring statistical inference, modelling, hypothesis testing, or replication, the underlying dataset should be cited directly rather than the narrative report. The dataset used in this report is openly published by the Human Clarity Institute and includes full variable definitions, methodological notes, and supporting documentation suitable for secondary analysis.
Suggested citation (report):
Human Clarity Institute. (2026). The Loneliness Trade-Off: AI Companions and the Rise of Synthetic Intimacy.
Suggested citation (dataset):
Human Clarity Institute. (2025). AI Companionship & Human Connection Survey 2025 [dataset].
Readers seeking deeper understanding may explore other Human Clarity Institute reports examining related topics, including digital fatigue, attention, trust in AI systems, and values alignment in digital environments. Together these reports form part of the Institute’s ongoing research programme investigating how emerging technologies influence human cognition, behaviour, and wellbeing.
Researchers seeking technical detail, replication, or extended analysis are encouraged to consult the underlying dataset directly through the Human Clarity Institute’s open data library, where full documentation, variable definitions, and supporting materials are available for further study.
© 2026 Human Clarity Institute. All rights reserved.
Values vs Noise
Values vs Noise explores how digitally saturated environments compete with attention, reflection, and internally guided priorities — shaping coherence, direction, and meaningful human functioning over time.
Human Values in Practice
Human Values in Practice examines how values alignment, direction, attention, and behavioural drift interact to shape coherence, self-trust, and human functioning within digitally mediated environments.