Two Heads Are Better Than One Redux: AI, OR/MS and Society

The Monday plenary session at the 2025 INFORMS Annual Meeting in Atlanta began by recognizing 2025 INFORMS awards and prizes winners, and expressing gratitude to the sponsors who made the conference possible. This opening gesture set a tone of appreciation and unity, acknowledging both individual achievements and collective support. Following the introduction, the plenary speaker, Ramayya Krishnan, delivered an insightful address that explored the evolving partnership between artificial intelligence (AI), operations research and management science (OR/MS), and the critical role these disciplines play in shaping a responsible, effective and human-centered future for technology.

Krishnan emphasized that the convergence of AI and OR/MS is more important today than ever before, particularly as society transitions from isolated AI capabilities toward system-level outcomes that demand integration, optimization and governance. He illustrated how OR/MS provides the analytical backbone that can help operationalize AI, transforming isolated algorithms into scalable, reliable systems that serve real societal needs.

One of the most compelling sections of his talk focused on the practical applications of AI in education and industry. Krishnan highlighted recent innovations such as live translation technologies, AI-supported academic collaboration and generative AI tools that now improve technical documentation and coding workflows by 25%-30%. He noted examples like AlphaEvolve, which has discovered new heuristic scheduling solutions, and AI systems achieving gold-medal performance in international mathematics competitions – a striking indication of how AI is not just automating, but actively advancing human problem-solving.

However, he cautioned that these advances represent only the “tip of the iceberg.” Beneath the surface lies deep societal and structural challenges, framed through the 70-20-10 principle, in which success depends on the right balance among people, technology and AI. Krishnan warned of the potential consequences for “exposed occupations,” in which automation could reduce job availability, and highlighted ongoing debates around intellectual property rights in AI-generated work. In response, he referenced America’s AI Action Plan, which calls for the creation of AI ecosystems that foster innovation while ensuring fairness, transparency and inclusion.

A recurring theme in Krishnan’s address was the idea of moving from capability to system-level outcomes – a transformation that requires the OR/MS toolkit. He presented three central questions that researchers and policymakers must consider:

1. O.R. for AI – How can operations research study and improve AI systems as analytical objects?
2. Operationalizing AI – How can we deploy AI solutions efficiently and ethically across industries?
3. Governance and Workforce Transition – How can institutions support workers as AI reshapes career paths and skills?

Krishnan supported these questions with multiple case studies. The first explored multi-turn conversational AI, analyzing how digital agents engage users through questioning, acting and expressing. He described how researchers use “adversarial prompts” to stress-test model robustness, creating digital proving grounds to measure how well AI systems survive under challenge. This “survival of the model” framework helps identify points of performance drift and failure, enabling the design of more resilient algorithms.

Another case study examined ambient AI in healthcare workflows. Krishnan described how physicians now conduct consultations while AI systems automatically transcribe and summarize conversations, integrating them into electronic health records. These tools enhance productivity and accuracy but also raise governance and ethical questions. For instance, as transcription roles become automated, how do we ensure equitable career transitions and provide reskilling pathways for displaced workers?

Drawing on findings from a 2024 research paper, Krishnan discussed AI exposure across occupations, showing that roughly 18% of professions face at least 50% task overlap with AI capabilities. However, he also highlighted an opportunity: Many individuals possess transferable skills that could enable them to shift into emerging fields, if supported by proper education and guidance. His concept of “opportunity landscapes” illustrates how AI and OR/MS can map human talent more effectively, revealing hidden compatibilities and new training routes. For example, studies show that 45% of individuals who are strong candidates for alternative roles remain unaware of their potential fit. Through intelligent systems like SkillFit, AI can help people identify and pursue new paths, turning automation risk into social mobility.

Krishnan concluded by calling for the creation of “digital land-grant universities,” institutions positioned to deliver AI-driven education to communities everywhere. Such universities, he argued, could democratize access to skills, create industry-ready graduates and ensure that AI’s benefits are distributed equitably across society.