Negotiating with technology
Working with AI should be an iterative process in which responsibility, creativity, and democratic participation guide how the technology is regulated, co-constructed, and deployed.
At universities and companies adopting generative AI systems, we humans no longer simply accept or reject the tool. We adjust prompts, check results, set limits, and sometimes rewrite the rules of use. The paper by Priyadharshini and Shantichitra argues that this everyday practice amounts to an active negotiation between algorithmic capability and human values. To describe it, the authors propose what they call a human-AI negotiation framework, organized around three values: responsibility, creativity, and democratic participation. Those values give the negotiation its structure and displace the traditional view of "adopters vs. laggards."
The core argument starts from the shortcomings of classic technology adoption models, which assume one-off, static decisions. AI, by contrast, keeps adapting, and people shape it by interpreting and modifying it, both individually and as organizations. Drawing on science and technology studies, critical AI ethics, and agency theory, the paper shows that interacting with AI is an iterative process in which these values guide how the technology is regulated, co-constructed, and deployed.
For people who run organizations, this means planning AI projects (and AI use) as ongoing conversations rather than one-way rollouts. That requires feedback channels where users can raise concerns, propose creative uses, and take part in decisions on governance policy. Training, ethics committees, and audit mechanisms need to be designed around those three dimensions, not only to measure adoption speed.
The word negotiation, though, implies two parties that can each give ground. Whoever writes a prompt can insist, correct, or drop the tool. What they cannot do is touch the model, its data, or its built-in limits, which are decided far from the organization using it. The negotiation the paper describes takes place, then, within a margin drawn by others. That margin exists and is worth occupying, but mistaking it for an agreement between equals is the most comfortable way to delegate without noticing.
Conclusions
- Organizations need to create institutional spaces where people discuss and adjust how they use the technology.
- Regulatory frameworks need to recognize human agency and allow people to keep reinterpreting the systems.
- Training should focus on negotiation and critical skills, not only on technical operation.
- Impact assessments should measure how well outcomes align with the values of responsibility, creativity, and democracy.
Source: From AI adoption to human - AI negotiation, Priyadharshini K, K. Shantichitra, Journal of International Students, 2026.