How People Experience Autonomy and Control in AI-Mediated Decisions — 2026 Data
This page summarises findings from the Human Clarity Institute’s Autonomy, Control & Perceived Independence 2026 dataset, based on 352 valid responses across six English-speaking countries.
The research examines how people experience control when interacting with AI systems, whether they feel influenced or steered, and where they draw boundaries around intervention and autonomy.
Within the broader Human–AI decision system, this data focuses on how people experience control, independence, and the ability to intervene when using AI in decisions.
See the full system model:
A data-driven model of how AI reshapes thinking, confidence, control, and decision behaviour
View the Autonomy, Control & Perceived Independence 2026 Dataset
What the data shows
People often use AI while still experiencing themselves as in control of their decisions, even when those systems influence how choices are evaluated.
Four signals stand out in this dataset:
Most people report feeling in control when using AI. Many also experience nudging or subtle steering. Confidence in overriding systems is very high, and regret following AI-influenced decisions is relatively uncommon.
Together, these findings suggest that perceived control remains strong, even while influence is present.
Across HCI’s wider behavioural datasets, this pattern increasingly suggests that people often attempt to preserve agency and judgement ownership even while adapting to AI-supported environments.
People often experience themselves as in control while still adjusting their decisions in response to system input.
Feel in control when using AI
Most people report a strong sense of control over their decisions when interacting with AI systems.
People generally feel able to make their own decisions even when AI is involved.
Feel nudged or steered by systems
At the same time, many report experiencing subtle influence or directional nudging in AI-supported environments.
People often notice that systems influence how they think about decisions, even when they do not fully rely on them.
Say they can override AI recommendations
Confidence in the ability to ignore or override system suggestions remains very high.
People typically feel able to ignore or override system recommendations when they disagree.
Regret AI-influenced decisions
Only a small share report frequent regret, suggesting that most do not experience strong negative outcomes from AI-supported choices.
Most people do not frequently experience negative outcomes after using AI in decisions.
Taken together, these findings point to a consistent tension: people report feeling in control and capable of overriding systems, while also recognising that those systems can shape behaviour and decision direction. Perceived autonomy remains intact, but influence is visible. This combination of perceived control and recognised influence reflects a broader pattern in which AI shapes how decisions are approached, while people continue to experience themselves as the final authority. In practice, perceived autonomy appears to be maintained less through complete independence from systems and more through the continued ability to intervene, override, and reassert personal judgement when needed.
By the numbers (from HCI data)
Default to making the decision themselves when uncertain
When unsure about an AI recommendation, many prefer to intervene and retain direct control rather than accept the system output.
Among those who feel nudged, many still report being in control
Perceived influence does not eliminate perceived autonomy, suggesting that people can feel both guided and in control at the same time.
Several behavioural patterns discussed on this page — including cognitive strain, behavioural reliance, delegated judgement, verification behaviour, and decisional uncertainty — were already documented within pre-generative-AI research on human decision-making. View the historical baseline
Patterns observed in the data
Control and influence coexist
The clearest pattern in this dataset is that perceived control and perceived influence are not opposites. Most people report feeling in control, yet many also recognise that systems can nudge or steer their behaviour. This helps explain why people may adjust or reassess their thinking when using AI, even while maintaining a strong sense of personal control.
This coexistence of influence and perceived autonomy is one of the clearest behavioural signals in AI-assisted decision-making, particularly where people continue to experience themselves as responsible for the final outcome.
Perceived control does not depend on constant intervention
Even though most feel capable of overriding AI, this does not always translate into frequent action. Control appears to be experienced as a latent ability rather than something that must be exercised continuously.
In behavioural terms, this suggests that many people experience agency less as constant resistance and more as confidence that intervention remains available when it matters.
Boundaries are situational, not fixed
Intervention thresholds vary depending on context. When uncertain, many default to making the decision themselves, suggesting that perceived autonomy is maintained through selective intervention rather than constant resistance.
Influence does not necessarily produce negative outcomes
Despite widespread perceptions of nudging, regret remains low. This suggests that influence is often experienced as subtle or acceptable rather than harmful or coercive.
In practice, system input is usually incorporated into decisions rather than replacing personal judgement, reflecting how people maintain a sense of control while interacting with AI systems.
Methodology
This dataset forms part of the Human Clarity Institute’s Human–AI Experience research programme, examining how people experience autonomy and control when interacting with AI systems, how influence and nudging are perceived, and where boundaries around intervention, override, and decision ownership emerge. The study uses a cross-sectional online survey design and focuses on descriptive patterns in perceived control, nudging awareness, override confidence, regret following AI-influenced decisions, and decision intervention thresholds.
Data were collected via the Prolific research platform from adults across six English-speaking countries. Participants provided explicit consent for anonymised open publication as part of HCI’s open research programme.
Sampling & participants
- Final n: 352
- Countries: United Kingdom, United States, Canada, Australia, New Zealand, Ireland
- Eligibility: Adults aged 18+ from six English-speaking countries
- Recruitment platform: Prolific
The resulting dataset should be interpreted as a non-probability convenience sample and is not intended to represent national populations.
The cleaned dataset, variable dictionary, and reuse terms are publicly available through the HCI dataset repository: Autonomy, Control & Perceived Independence 2026 Dataset →
Data integrity
All percentages reported on this page are calculated from valid responses in the cleaned dataset (n = 352). Percentages are rounded to the nearest whole number for readability. Unless otherwise stated, summary percentages combine respondents selecting 5–7 on the 7-point agreement scale (slightly agree, moderately agree, or strongly agree).
Where percentages refer to subgroups or intervention conditions (such as uncertainty or decision thresholds), the wording on the page makes that explicit. Comparative patterns reflect positioning within the relevant subgroup rather than across the full sample.
Participant IDs, timestamps, and direct identifiers were removed before publication as part of the anonymisation process.
This dataset is exploratory and descriptive in nature. It does not support causal inference and results should be interpreted as observed patterns within the survey sample.
This dataset is released as open research to support transparent analysis of perceived control, AI influence, intervention boundaries, and the human experience of maintaining autonomy in digitally mediated decision environments.