In late 2024, we hosted a cutting-edge academic webinar that brought together experts at the intersection of artificial intelligence and optimization.
Titled “The Impact and Uses of Generative AI in Optimization: What’s Next,” the session featured Jerry Yurchisin, Senior Data Science Strategist at Gurobi; Thomas Braam, Senior DevOps Engineer at Gurobi; Dr. Warren Hearnes, Founder of OptiML AI; Dr. Can Li, Assistant Professor at Purdue University; and Dr. Irv Lustig, Optimization Principal at Princeton Consultants.
As generative AI technologies rapidly evolve, our panelists discussed how these tools are beginning to influence the modeling, solving, and application of mathematical optimization.
While generative AI is often associated with image generation or text completion, the panelists discussed how it can also serve as a creative partner in optimization workflows. Tools like large language models (LLMs) are beginning to assist in everything from writing optimization code to identifying model structures and suggesting solution strategies.
Jerry Yurchisin opened the session by noting, “We’re seeing generative AI help users bridge the gap between business language and mathematical formulation. It’s like having a co-pilot for model development.”
This is especially valuable for users who are not optimization experts, but who need to make complex decisions based on constraints, trade-offs, and objectives.
One of the most interesting use cases we discussed involved converting natural language problem descriptions into mathematical optimization models. Dr. Can Li presented early research on how LLMs can help formulate decision variables, constraints, and objectives based on text input, allowing a broader range of users to access optimization tools.
“The goal is to make optimization more accessible,” said Dr. Li. “If someone can describe their problem in plain English, a generative AI model could help translate that into a structured model.”
While promising, this approach also raises concerns about correctness, model validation, and the need for human oversight.
Rather than fully automating optimization, panelists emphasized that generative AI should be seen as a collaborator. It can provide suggestions, generate boilerplate code, and even offer novel approaches. But it’s up to the human expert to refine and verify the solution.
Gurobi’s Thomas Braam warned that “we must be cautious about over-reliance on AI-generated outputs.” He stressed the importance of building guardrails to ensure that models are both correct and ethical.
This hybrid approach—AI for exploration, humans for validation—reflects the current best practice in fields like data science and machine learning and is likely to guide the evolution of optimization workflows as well.
The panel also addressed the ethical dimensions of AI-assisted optimization. Dr. Warren Hearnes raised concerns about transparency and fairness in AI-influenced decision-making. “We need to make sure these tools support compliance and equity, especially in regulated or sensitive environments.”
The group agreed that explainability is key. If a model’s logic or output isn’t clear, users and stakeholders may not trust or adopt it. Generative AI must support, rather than obscure, understanding.
Dr. Irv Lustig noted that no matter how powerful generative AI becomes, it won’t replace the core value of optimization. “You still need a solver to find the best solution. AI can suggest models or explore problem variations, but it can’t do what an optimization engine like Gurobi does.”
In fact, the synergy between generative AI and solvers like Gurobi could lead to more efficient development, better prototyping, and faster innovation.
Looking forward, the panelists predicted that AI-powered interfaces and low-code platforms will become more prevalent. These tools will make it easier for business users, analysts, and students to interact with optimization without the need for deep technical expertise.
The discussion also highlighted the importance of education. As optimization and AI converge, training the next generation of professionals to use generative AI both responsibly and effectively will be critical.
The full recording of ”The Impact and Uses of Generative AI in Optimization: What’s Next?” is available on-demand here. It offers valuable perspectives for educators, researchers, and practitioners exploring how AI is reshaping the way we solve complex problems.
Gurobi remains committed to supporting innovation at the intersection of AI and optimization. By empowering researchers and developers with world-class tools, we are helping to shape the future of intelligent decision-making. Learn more about our AI-assisted tools for optimization here.
Choose the evaluation license that fits you best, and start working with our Expert Team for technical guidance and support.
Request free trial hours, so you can see how quickly and easily a model can be solved on the cloud.