When Energy Secretary Chris Wright visited New Mexico last February, he described artificial intelligence as having the same urgency as the Manhattan Project. Then in November, President Trump signed an executive order launching DOE’s Genesis Mission to advance the technology for government work.

The White House’s commitment is clear. But how much of this AI fervor has made its way through the rank and file of the Labs?

Sandia polled members of its workforce in August with a single question, asking their feelings about using the technology at work. An astounding 8,573 people responded, 79% saying they saw at least some benefit to AI. The Lab News asked AI leaders from around the Labs for their reactions to the poll results, and about trends, opportunities and challenges they see surrounding AI in our workplace.

Tina Nenoff, Sandia Fellow, Labs AI strategy lead

I think positive reinforcement from leadership at all levels to the staff is helping drive the strong response to AI at the Labs. Adding top cover from management to staff for the incorporation of AI into mission work is vital, as is covering risks that might arise for bringing this new technology into the workflow.

Sandia’s leadership is fully committed to incorporating AI into the Sandia enterprise in a safe and secure manner. To help teams scale up AI workflows into mission-critical work, Sandia’s Senior Leadership Team has made an AI investment area in FY 26 that is working toward a minimum viable capability that includes Sandia’s inclusion into the Venado/OpenAI collaboration venture at LANL, the Sandia toolkit and SandiaAI Chat.

Furthermore, Sandia’s AI Board of Directors and our leadership teams for the DOE-sponsored Genesis work and the NNSA-sponsored AI for National Security program are working to introduce AI research and development into mission work through demonstration projects, use cases and lighthouse problems, especially difficult tasks that fold back into the Genesis Mission.

I often tout that Sandia is the premier engineering lab for national security. But I think in just a couple of years, when we’ve fully incorporated AI into our mission workflow, that designation will be unquestioned.

Dan Turner, AI for Nuclear Deterrence lead

While these are positive poll results, what would be more interesting would be the number of people who could testify to transformational impacts from AI.

In some cases, I’ve seen the initial excitement over what AI could do subside as people realize how hard it is to achieve transformational results. Many pilot projects are showing progress, but they struggle to scale when the pilot is over. What we need are sustained efforts that can overcome these challenges by leveraging the tools we have in creative and effective ways.

One way Sandia is moving through this challenge is through our involvement with Genesis Mission, the large DOE investment in AI. This will provide the AI platform and models we need to scale to an enterprise AI capability.

However, it all boils down to focusing on what will lead to the highest return on investment. There are lots and lots of things that could be done based on how interesting they are from a technical perspective, but the efforts that will win the day will be those that pay for themselves and then some.

Gregory Butler, Facilities and Infrastructure AI champion

In a time when we’re all being asked to do more with less, AI, when used correctly, is proving to be a valuable tool.

In Facilities, we utilize a containerized chatbot with a visual language model to extract and analyze fire protection data from scanned forms, which helps to eliminate a considerable amount of administrative time.

We have plans to take advantage of chatbots even more in the future as retrieval-augmented generation, or RAG, tools improve. There’s a lot more we’d like to do, but there are a lot of moving parts we have to consider before we can jump into them.

When determining what role AI should have in our processes, our goal is to start with asking, “What is the problem we are trying to solve?” In some cases, we’ll discover that an AI tool is what could best meet the need. In others, we may discover there’s another IT solution or process update that could meet the need even better. We want to avoid starting with the cool AI tool and just trying to think of where we can use it.

Siva Rajamanickam, BANYAN Generative AI Institute

I am an AI optimist, and I am not surprised by the 79% number. I would like that number to be 100%.

In terms of computing research, we are focused on approaches that are critical for our mission. For example, learning how to integrate AI agents with our modeling and simulation tools, experimental capabilities, data sets, prototyping and manufacturing capabilities are all important research problems.

We are also starting to explore federated training and inference approaches at the tri-labs and at the complex level. We are working toward building an AI platform that will become a foundation to transform the future of science and security.

AI will allow us to do our jobs far more efficiently than we have ever done. I also believe the way we will do our work will be vastly different. Imagine being a designer and having a suite of customized AI agents that can help create new questions, generate hypotheses, generate a set of designs that match the hypotheses, evaluate those designs and present our designers with multiple choices over a weekend.

This might sound like Iron Man and JARVIS, but it is not too far-fetched.