OP-ED: Newfoundland and Labrador’s AI Procurement Reforms Reveal a New Governance Challenge
A recent consulting report prepared for the Newfoundland and Labrador government has raised significant concerns regarding the use of artificial intelligence in research. The report included citations that did not exist, prompting scrutiny and highlighting the need for improved governance in public procurement.
In response, the government has tightened procurement rules, requiring vendors to disclose whether AI tools are used in preparing reports and allowing officials to audit that usage. This situation reveals a deeper issue: public procurement systems still evaluate expertise using credibility signals designed for a pre-AI world.
The Evolution of Knowledge Production
With the introduction of generative AI, the model of knowledge production has changed. Human judgment and machine synthesis now work together, complicating how expertise is assessed. When governments commission major studies, they purchase not only strategic advice but also institutional credibility. Traditionally, a report’s authority relies on visible signals of methodological rigor such as literature reviews, extensive reference lists, and detailed documentation of sources. These elements help public officials determine whether a report’s conclusions are based on serious research or speculation.
However, the integration of AI-assisted synthesis into the research workflow means that these signals only provide a partial view of how analyses are conducted. Two reports may appear identical, but one may be the result of careful human verification while the other heavily relies on automated synthesis with limited oversight.
Challenges in Procurement Systems
Procurement systems initially designed for human research struggle to differentiate between these processes. While artificial intelligence does not eliminate expertise, it alters the way expertise is produced and verified. The government of Newfoundland’s response is crucial in this context.
By mandating that vendors disclose the use of AI tools and allowing audits, the province is introducing transparency into the research process funded by public money. This reform enables a closer examination of the analytical methods behind reports, even when machine-assisted tools are involved.
A Broader Implication for Governance
Newfoundland’s procurement reforms represent an early institutional response to a broader shift in expertise production and evaluation. Few governments have started to adapt their oversight mechanisms to account for the reality that AI tools are already integrated into research and advisory work across various sectors, including legal analysis, academic research, and corporate strategy.
Governments across Canada spend billions annually on consulting studies that influence critical decisions in areas such as health care staffing, infrastructure planning, and regulatory policy. As AI tools become more prevalent in these processes, procurement systems nationwide will face the same challenges as Newfoundland: evaluating not only the conclusions of reports but also the methodologies that generate expertise.
Effective Oversight Measures
To achieve effective oversight, clearer disclosure rules, stronger expectations for human verification of sources, and the capacity to audit analytical work are essential. These measures will allow new tools to enhance research capabilities while maintaining confidence in the expert advice that shapes public policy.
The recent controversy surrounding a Newfoundland and Labrador government report may ultimately be less significant for the errors it revealed than for the governance questions it raises. As artificial intelligence continues to integrate into research and analytical functions across public and private sectors, institutions must evolve their mechanisms for evaluating expertise.
Newfoundland’s procurement changes indicate that this adaptation is underway. Systems designed for the twentieth century must now evolve to govern a new model of expertise where human judgment and machine synthesis increasingly collaborate.
