A recent research study published at the CHI '26 conference, conducted by researchers from the University of Southern California and IBM Research, has shown a fundamental discrepancy in perspectives between freelancers and their clients regarding the use of artificial intelligence tools in completing professional tasks.
The Reality of AI Usage in the Freelance Market
The findings indicate that AI has become an integral part of the freelance workflow, with 78.8% of freelancers reporting the use of these technologies to keep up with deadline pressures and price competition. On the other hand, 81.3% of clients expressed encouragement for the use of AI in one form or another, reflecting an initial acceptance of this technology by both parties.
Why is Disclosure Absent?
Despite this acceptance, the study revealed a gap in transparency, as only 8.6% of freelancers voluntarily disclose their use of AI. This phenomenon is attributed to several reasons:
- Considered a Standard Tool: 53.8% of freelancers believe that AI is no different from traditional office tools like Microsoft Office, and therefore does not require reporting.
- Implicit Assumption: More than 60% of freelancers believe that clients are capable of detecting AI usage in deliverables, and they consider the client's lack of objection as implicit approval.
- Lack of Oversight: Clients lack confidence in their ability to detect AI usage, even though nearly a quarter of them demand explicit disclosure.
The Conflict Over Task Nature
The study highlighted a conflict in the classification of tasks between "assistive" (which can be automated) and "core" (which require purely human effort). While both parties agree that proofreading is an assistive task, they strongly disagree on tasks such as writing emails, research, and data analysis, where freelancers tend to classify them as assistive, while clients view them as core tasks that should not be delegated to a machine.
Do Usage Policies Solve the Problem?
The researchers tested the impact of establishing clear "usage policies" and found that the results were unexpected; freelancers tend to interpret restrictive rules more broadly than intended, while they interpret encouraging rules more conservatively, suggesting that simply setting rules does not end the uncertainty in the freelance work environment.