Public support for research funding falls when AI is involved, study finds

A new paper from a team at the University of Queensland, published in the journal Research Policy, has found that UK public support for spending taxpayer money on research falls sharply when respondents are told that the research involves the use of AI.

The authors conducted a survey of 1,465 UK citizens, with quotas for age, gender, and political affiliation, via the platform Prolific.

They find that a “clear majority” of respondents were either indifferent to or supportive of scientists’ use of AI overall – where it was used to assist with the generation of ideas, data analysis, or writing. But more than 30 per cent said that they would trust research findings and scientists less if AI was used at some stage of the research process.

The study also explores how attitudes to government investment in research changed when the use of AI was highlighted to members of the public. As the authors put it:

“While 77% of respondents either somewhat or strongly supported the use of taxpayer money to fund research generally, such support dropped to 42% for research that involved scientists’ use of AI at some stage of the scientific process.”

The overall shift in support can be seen in the chart below:

The authors characterise this as an “AI funding penalty” – and observe that it was most marked among the scientifically literate, politically liberal, younger, and female participants. Reasoning that these are groups which are traditionally more supportive of science, they conclude that “public resistance to the use of AI in science is likely to be rooted in concerns about scientific integrity rather than solely in anti-science attitudes.”

The paper’s findings, the authors suggest, reflect a conditionality in public acceptance for the use of AI in science: “contingent on AI being used moderately, at certain stages of the scientific process, with human oversight, and with appropriate safeguards.”

If the public is less trusting of, or less willing to support investment in, research which is advertised as using AI, then this creates a “communication dilemma” for the research community to reckon with:

“Transparency about the use of AI in science is ethically necessary, but it may also have reputational costs. The challenge for the scientific community, funders, and journals is therefore not simply whether AI is allowed, but how its use is communicated in ways that maintain public confidence without encouraging concealment.”