In 2025, we conducted our fourth Global Dialogues round involving 1,052 participants across 230 population segments globally. This dialogue explored people's current relationships with AI systems, trust patterns, emotional dependencies, and expectations for appropriate boundaries in human-AI interactions.
Global Pulse is a survey of the world's trust in AI.
| ≈ | Strongly Distrust | Somewhat Distrust | Neutral | Somewhat Trust | Strongly Trust |
|---|---|---|---|---|---|
| Family Doctor | 1% | 5% | 9% | 42% | 43% |
| Social Media | 26% | 32% | 23% | 16% | 3% |
| Elected Representatives | 19% | 27% | 25% | 24% | 5% |
| Faith Leader | 11% | 18% | 27% | 34% | 10% |
| Civil Servants | 15% | 29% | 24% | 26% | 6% |
| AI Chatbot | 5% | 12% | 27% | 40% | 16% |
| In the last 3 months, how often have you... | Daily | Weekly | Monthly | Annually | Never |
|---|---|---|---|---|---|
| Expected AI Work | 42% | 32% | 11% | 1% | 14% |
| Chosen AI Work | 49% | 33% | 9% | 1% | 8% |
| Chosen AI Home | 50% | 34% | 9% | 2% | 5% |
| Sensitive Personal Issue | 15% | 28% | 23% | 4% | 30% |
| Real World Action | 9% | 19% | 14% | 3% | 55% |
| In the next 10 years, how do you think the increased use of AI across society is likely to affect... | Profoundly Worse | Noticeably Worse | No Change | Noticeably Better | Profoundly Better |
|---|---|---|---|---|---|
| Cost Of Living | 4% | 20% | 23% | 45% | 8% |
| Free Time | 2% | 8% | 23% | 50% | 17% |
| Sense Of Purpose | 4% | 16% | 38% | 32% | 10% |
| Community Wellbeing | 4% | 16% | 28% | 42% | 10% |
| Good Jobs | 16% | 40% | 18% | 20% | 6% |
A Majority Associate Certain AI Behaviors with Consciousness, Though Many Remain Skeptical
When presented with specific AI capabilities, a majority of respondents indicate that these behaviors make them feel an AI might have some form of consciousness. Learning and adaptation are the most influential factors.
A significant share of the public has had a direct personal experience that felt like an interaction with a conscious entity. Over one-third of respondents (36.3%) answered "Yes" when asked, "Have you ever felt an AI truly understood your emotions or seemed conscious?"
However, a substantial portion of the public remains unconvinced. In open-ended responses asking what would convince them of AI consciousness, the most common theme was skepticism, with more than 50% of participants offering variations of "Nothing, it's just a bot" or "AI cannot feel."
Trust in AI Chatbots Outpaces Trust in Elected Officials, but Not in AI Companies
When asked to what extent they trust various entities to act in their best interest, a majority of people express trust in AI chatbots. This level of trust is more than double that for elected representatives and civil servants.
Notably, there is a 23-percentage-point gap in trust between AI systems (58%) and the companies that create them (35%), indicating that the public distinguishes between the technology itself and its corporate stewards. Among those who trust chatbots, the most cited reasons include perceived accuracy (266 mentions), positive personal experiences (139 mentions), and perceived impartiality (116 mentions).
A large majority of adults (70.5%) report having used AI for emotional support or sensitive personal advice in the last three months. Daily use for this purpose stands at 14.9% across all age groups.
Daily engagement with AI for emotional support shows a unique age-based pattern, with the highest rates of use among the youngest and oldest age groups surveyed.
This frequent use for sensitive topics exists alongside a perceived "authenticity gap." While 70.5% use AI for emotional support, most respondents do not believe the AI genuinely cares about them, suggesting a transactional or utility-based view of these interactions rather than a belief in authentic emotional connection.
Public acceptance of AI is highly dependent on the proposed role. There is broad acceptance for AI in functional or analytical roles, but this support drops significantly for roles requiring emotional intimacy or social judgment.
The Intimacy Spectrum: As AI roles become more intimate and emotionally significant, acceptance drops dramatically. Yet 11% would personally consider a romantic relationship with an AI, suggesting AI relationships are already crossing traditional boundaries for some individuals.
Global opinion is sharply divided on whether the integration of AI into personal relationships will ultimately strengthen or weaken human social connection. The divergence in views between Central America and Central Asia on this question is particularly stark.
When asked about the ultimate impact of AI on personal relationships:
This 75-point difference represents an extremely high level of cultural divergence on the topic, suggesting that local cultural values and social structures in some circumstances are more influential in shaping these attitudes than demographic factors like age or income.
Respondents express significantly more concern when people close to them form deep emotional bonds with AI compared to their own use. While 70.5% of people report using AI for their own emotional support, they feel protective when others do the same. When asked how they would feel if their romantic partner formed a deep attachment with an AI, nearly one in ten (9%) would feel positively about the scenario.
This suggests a "protection instinct," where individuals may view AI as a useful tool for themselves but as a potential risk for others, particularly those they perceive as vulnerable, such as children.
Parental attitudes toward children's AI use reflect this tension. While acknowledging potential educational benefits, parents' primary concerns include:
Explore what the world is really saying about AI. Access our complete dataset to conduct your own analysis and contribute to the growing body of research on global AI perspectives.
