AI Use in Award Nominations

When an award nomination is written with the help of artificial intelligence, what exactly should the award committee judge? 

That question prompted our INFORMS Taskforce, at the request of the Professional Recognition Committee, to consider how generative artificial intelligence should be handled in the nomination process. Generative AI is software that can produce or revise text in response to a user’s instructions. It can help organize evidence, improve clarity, and turn rough notes into polished prose. It can also introduce errors, exaggerate claims, or flatten a distinctive professional record into generic praise. 

Those competing possibilities have led some organizations to consider requiring disclosure or limiting AI use in professional submissions. We take a different view. Nominators should be responsible for the accuracy and integrity of what they submit. Award committees should judge the nominee’s contributions and impact, not the tools used to prepare the nomination. 

A Long Tradition of Professional Recognition 

INFORMS has a long tradition of professional recognition. ORSA established the Frederick W. Lanchester Prize in 1954, TIMS created the Franz Edelman Award for practice in 1972, and the two societies jointly established the John von Neumann Theory Prize in 1975. After the 1995 merger, INFORMS continued to expand its recognition programs. New honors included the Saul Gass Expository Writing Award in 1996, the Daniel H. Wagner Prize in 1998, the Impact Prize in 2004, the UPS George D. Smith Prize in 2012, the Donald P. Gaver, Jr., Early Career Award in 2020, and the Early Career Practitioner Award in 2024. 

Today, INFORMS lists 25 institute-wide prizes and awards, while its communities confer another 176 recognitions. In all, the INFORMS community offers more than 200 distinct recognitions. 

At that scale, nomination policy shapes how the profession identifies and honors achievement. 

What This Guidance Covers 

Our guidance addresses the nomination materials themselves, including the letters, statements, and supporting materials prepared to make the case for an award. It applies equally to nominations submitted by others and to self-nominations. 

This guidance does not address the use of AI in the work being recognized. That raises a much broader set of questions. Once we begin judging AI-assisted writing, it is difficult to avoid questions about AI-assisted analysis and, more generally, about the computational tools our field has relied on. Consider a proof that holds because a computer verified 128 separate cases. Where should the line be drawn between acceptable assistance and work that is no longer meaningfully the researcher’s own? 

Those are important questions, but they are not really questions about award nominations. They involve authorship, research methods, and the standards that should apply to different types of scholarly work. In our view, they also go beyond the mandate of a two-person taskforce. They deserve separate consideration and broader input from the INFORMS community. 

How Other Institutions Approach AI 

Other institutions have also confronted aspects of AI use that are directly relevant to nomination policy. Their policies reveal that restrictions tend to be strongest when confidentiality is at stake. Uploading confidential material to an AI system can expose information to a third party and compromise confidentiality. Accordingly, NIH bars peer reviewers from using generative AI to analyze applications, while NSF similarly restricts reviewers from placing proposal material into unapproved tools. 

Disclosure is more common when AI assists in producing the scholarly work itself. Elsevier and Nature require authors to disclose relevant AI assistance, while the International Committee of Medical Journal Editors recommends that journals require such disclosure. The Leiden Declaration on Artificial Intelligence and Mathematics similarly calls for disclosure while insisting that responsibility for correctness remains with human authors. 

INFORMS has adopted a related principle in publishing. Management Science treats submission as a declaration that authors have reviewed and edited any AI-generated content as needed and accept responsibility for its accuracy. 

An award nomination is different from both peer review and scholarly authorship. It is neither someone else's confidential work entrusted to a reviewer nor the underlying contribution being honored. It therefore warrants its own standard. 

What Award Committees Actually Evaluate 

Professional recognition has always depended on more than the work itself. Someone must identify a deserving colleague, assemble evidence, explain the significance of the contribution, and persuade a committee that the record meets the award’s standards. That process requires professional judgment, but it also requires time and writing skill. 

A nomination letter is important to that process, but it is not the achievement being honored. A committee considering a research prize should evaluate the research, the advancement it represents, and evidence of its influence. A committee considering a practice award should examine the results achieved, the organizations or communities affected, and the nominee’s role in producing those results. The nomination serves as a guide to that record; it does not substitute for it. 

This distinction matters because skillful writing cannot create merit where the underlying record does not support it or rescue an otherwise weak case. If AI helps the nominator present relevant evidence more clearly, the committee receives a more readable submission. If it produces vague superlatives, unsupported claims, or invented facts, the nomination should fail for the same reason as any other weak nomination: it does not establish that the nominee deserves the honor. 

The standard should remain the quality and credibility of the case, not the method used to communicate it.

Human Accountability Must Remain Central  

Permitting AI assistance does not mean treating its output as reliable. AI can misstate facts, invent references, blur distinctions, and reproduce biases. It may confidently attribute achievements to the wrong person or overstate the importance of a result. These are serious risks in any process that help define a profession’s historical record. 

The answer, however, is not to pretend that AI can be excluded from the nomination process. It is to make human accountability unmistakable. 

Every nominator should remain responsible for each factual statement, quotation, statistic, and characterization in the final nomination packet. The nominator should personally confirm that the evidence is accurate and that the assessment reflects his or her genuine professional judgment. A person should not sign a nomination that says more than the evidence supports or that makes claims the signer does not understand. 

This obligation also protects authenticity. The strongest nominations explain why a particular contribution mattered at a particular moment. They reflect knowledge of the field, the nominee’s role, and the obstacles that had to be overcome. AI can help structure that story, but it cannot supply the perspective of a colleague who truly understands the work. A generic letter may be grammatically flawless and still be unpersuasive. 

Confidentiality requires equal attention. Nomination packets can contain private letters, unpublished material, personal information, or proprietary business details. Nominators should not place such material into a third-party AI tool unless they are authorized to do so and understand how the tool stores and uses data. The principle is simple: sensitive information should not be shared with an unauthorized AI tool any more than with an unauthorized person. 

The Practical Limits of Mandatory Disclosure  

A disclosure rule may appear to offer transparency, but it would be difficult to apply fairly. AI can be used in many ways: correcting grammar, reorganizing paragraphs, generating an outline, summarizing notes, or drafting an entire letter. A policy would have to decide which uses require disclosure and how much assistance is enough to trigger the rule. 

Even then, enforcement would be unreliable. By watermarking submission PDFs with hidden instructions, organizers of the 2026 International Conference on Machine Learning detected 795 reviews that violated the conference's no-LLM policy. That technique catches only the most careless uses, is easily circumvented, and has no practical counterpart in award nominations. A disclosure policy would therefore depend largely on self-reporting. Conscientious nominators might disclose even limited assistance and risk having their submission viewed with suspicion, while others could remain silent without consequence. 

That would direct attention away from the nominee’s record and toward an essentially unprovable question about the submission’s composition. It could also disadvantage nominators who have less writing experience, write in a second language, or lack access to colleagues familiar with award conventions. The ability to produce a well-written nomination is not the same as the ability to recognize excellent work. 

In our experience, the challenge is often a shortage of strong nominations, not an excess. Highly deserving researchers, educators, practitioners, and volunteers are often overlooked because potential nominators lack time, confidence, or familiarity with the nomination process. Used responsibly, AI can lower those practical barriers without lowering the standard for professional recognition. 

A better approach is to require an affirmation that the nominator has verified the factual claims, has not disclosed confidential or other protected materials without authorization, whether through AI tools or otherwise, and stands behind the professional judgment expressed in the nomination packet. That creates a clear line of responsibility without turning AI use into a separate criterion. 

Nominations Are Not Peer Review 

Much of the concern about AI comes from rules developed for peer review, but the two activities are not the same. 

A peer reviewer receives confidential work from someone else and is entrusted to evaluate it independently. Uploading an unpublished manuscript or grant proposal to a third-party AI tool may violate that trust. Beyond confidentiality, the reviewer also occupies a neutral role in the decision-making process and should not delegate the evaluation to a machine. 

A nominator has a different role. The nominator is an advocate who assembles a case for recognition, usually based on published work, documented practice, or other established contributions. The award committee, not the nominator, remains the neutral evaluator. The committee must independently decide whether the evidence satisfies the award criteria. 

The same distinction places an obligation on award committees. Nomination materials are confidential, and the committee's evaluative judgment is not delegable. Peer-review rules should not automatically apply to nominators, but the reasoning behind those rules applies to the committees that evaluate nominations. 

A Practical Policy for INFORMS 

A workable policy can be both permissive and demanding. It should allow nominators to use AI for drafting, editing, and organization without requiring disclosure. At the same time, it should state clearly that the nominator is fully responsible for the final submission. The language below is offered as a starting point. It could be adopted, in whole or in part, in the guidelines that accompany INFORMS award submissions. 

INFORMS Guidelines for AI Use in Award Nominations  

• Permitted use. Nominators may use generative AI tools to help draft, edit, organize, or translate nomination materials. They do not need to disclose that use. 
• Responsibility. The nominator remains responsible for everything in the nomination. Facts, quotations, statistics, and descriptions must be checked, and professional judgments must be the nominator's own. 
• Evidence. Claims about a nominee should be supported by evidence the committee can verify. 
• Confidentiality. Nominators should not place confidential, unpublished, personal, proprietary, or otherwise protected information into an AI tool without permission. They should also understand how the tool stores and uses the information they provide. 
• Affirmation. At submission, the nominator affirms: “I have verified the factual claims in this nomination, I have not shared confidential or protected material without authorization, and I stand behind the professional judgments expressed here.” 
• Award committees. Nomination materials are confidential. Committee members should not put them into AI tools that retain the material or use it for model training. The committee must make its own judgments. 
• Misconduct. Fabricated references, false statements, and intentional misrepresentation are covered by existing INFORMS misconduct procedures, whether or not AI was used. 
• Review. This guidance reflects current AI capabilities and professional practice and should be reviewed periodically. 

This approach focuses on the integrity of the nomination. It does not reward or punish nominators for using a particular tool. It asks whether the evidence is accurate, whether confidential information was protected, whether the nominator stands behind the claims, and whether the committee exercised independent judgment. 

INFORMS should neither assume that AI can be kept outside the nomination process nor confuse assisted writing with outsourced judgment. An award nomination does not succeed on the strength of the prose alone. It succeeds because it accurately conveys the scale, originality, and impact of a person’s work through the informed judgment of a peer. 

AI may help communicate that judgment. It can never replace it.