The paper, called The Trust Gap, said that when asked about their confidence in AI output accuracy, 67% of respondents to a survey described themselves as “somewhat confident – still double-checking everything”, and a further 20% said results felt inconsistent.

Displayr said that 10% said they were very confident in the trustworthiness of AI and could easily verify and reproduce outputs.  

The research said that 71% of respondents fell into an ‘applying’ tier, which meant they used AI across specific workflows but had not yet moved to full integration, while 11% felt their AI use was embedded throughout their work.

Respondents also said that they were spending time manually re-checking outputs from AI, which was absorbing any efficiency gains from the technology.

A third ( 35%) said that their biggest desire was for AI that could marry speed and control, while 41% said that genuinely trustworthy AI could free them up to spend more time on insight work.

The findings are based on 203 responses to a survey fielded in March 2026.

Matilda Sarah, vice-president of sales and marketing at Displayr, said: “Researchers in this study understand data quality and methodological rigour, so we know the scrutiny they apply to AI outputs is the same scrutiny that makes their work credible to clients.

"What these researchers are describing is a verification gap. The workflows that close it – where outputs can be traced, reproduced, and stood behind professionally – are the ones that will define the next stage of AI adoption in the industry.”