A favorable view can have qualifications
Favorable and very favorable responses total 59.7%. The two unfavorable categories total 20.4%. Another 17.6% are indifferent and 2.3% are unsure. The chart retains all six categories so the neutral and uncertain responses remain visible.
Sentiment is a different measure from frequency, trust or accuracy. Someone can like the convenience of a tool and still review every answer. Someone else can use it often while remaining dissatisfied with parts of the experience.
Ask what sits behind an opinion
Citation's interpretation is that feedback becomes more actionable when it names an event. A comment such as 'the assistant helped me compare two explanations' points to a useful behavior. A comment such as 'the references did not support the paragraph' points to a specific failure.
A team can collect those examples alongside a sentiment question. This preserves the person's overall view while making it possible to decide what to improve. It also avoids treating every unfavorable response as resistance to change or every favorable response as proof of reliable output.
Keep different survey measures separate
When presenting results, label the question and its response base beside the percentage. Do not substitute the number of people who use a tool for the number who favor it. If two questions have different respondent counts, their results do not automatically describe the same set of people.
A trend comparison needs the same care. Wording, recruitment and category definitions can change between survey editions. The distribution on this page is a snapshot of the 2025 question; a claim about individual attitudes changing would require additional evidence.
How to read the data
This question has 33,412 responses from the All Respondents group. The six rounded shares sum to 100%. Combined favorable and unfavorable figures are calculated from the displayed values. The voluntary survey reflects its respondents and should not be treated as a random population sample.
The original report provides the published figures and question wording. The chart and downloadable data use the same values; labels are shortened for readability.
Questions about the findings
Why does the summary round favorable sentiment to 60%?
The chart retains one decimal place. Adding 22.9% and 36.8% gives 59.7%, which rounds to 60% when expressed as a whole percentage.
Does favorable sentiment imply trust in every answer?
No. The question asks about a stance toward AI in the workflow. Trust in accuracy is a separate question and a separate report.