SAP ran a quiet internal experiment to see how its experienced consultants judge work produced by artificial intelligence, and the outcome said more about people than about the software. According to an account published by VentureBeat in December 2025, the same AI-generated answers were praised when reviewers believed interns had written them and largely rejected when reviewers were told the truth.
How the test was set up
Five teams of consultants were asked to validate responses to more than a thousand business requirements. All of the responses had in fact been generated by Joule, SAP’s AI copilot, in a version built for consultants.
Four of the teams were told that the material came from junior interns fresh out of school. They worked through it, found it impressive and rated it roughly 95 percent accurate, VentureBeat reported.
The fifth team received identical content with one difference: they were told it had been produced by AI. That team rejected nearly all of it.
Second look, same answer
When the fifth team was then required to check each response individually, they found that the AI had not only produced correct answers but had also picked up on details that had previously been overlooked. Their final accuracy assessment landed at about the same figure as the other teams: roughly 95 percent.
The lesson SAP drew, as described in the report, is that resistance to AI inside organizations is often a matter of human attitudes rather than technical capability. Experienced consultants bring deep knowledge to their work, but they also bring biases about where good work can come from.
Positioning Joule as a support tool
SAP is presenting Joule for consultants as a way to strengthen human expertise rather than replace it. The company’s stated goal is to free consultants from time spent searching documentation and resolving technical detail so that they can focus on understanding the client’s business, its objectives and the decisions that follow.
According to SAP, junior consultants are becoming more independent with Joule’s help, while senior consultants are applying their experience to the harder problems where human judgment matters most. The company says this has created a new kind of collaboration between generations within consulting teams.
Next step: agents that understand processes
SAP’s longer-term plan is to move from copilots that respond to prompts toward what the industry calls agentic AI: systems that understand entire business processes and can propose solutions to complex problems with less human direction. The company points to its data on thousands of business processes as the foundation for that effort.
The experiment offers a caution for any organization rolling out such tools. If skilled staff dismiss accurate output simply because of its origin, the limiting factor on AI adoption will be trust, not performance.
