AI for the Arts: Lessons from the DeVos Institute's A³ Initiative

The Mark Morris Dance Group shares its experience participating in the DeVos Institute's A³ (Arts × Admin × AI) initiative, including key lessons for arts and cultural organizations considering AI adoption.

What is the DeVos Institute's A³ Initiative?

Artificial intelligence seems to dominate every conversation these days. Some see it as a game-changing productivity tool; others worry about its impact on creativity, jobs, ethics, and trust.

For arts organizations, those conversations can feel especially complicated.

Unlike many industries, arts organizations rely on deeply human work: creating art, building relationships, engaging audiences, cultivating donors, and serving communities. AI isn't going to replace those things; and it shouldn't. But could it help overstretched teams spend less time on repetitive administrative work and more time on mission-critical activities?

That's the question the DeVos Institute of Arts and Nonprofit Management set out to explore through its A³ (Arts × Admin × AI) Initiative, a multi-month research and pilot program involving sixteen arts and cultural organizations from across the United States. Rather than experimenting with AI for creativity's sake, participants focused on practical administrative challenges—from marketing and fundraising to operations, communications, and data analysis—to determine where AI could meaningfully improve capacity while remaining aligned with organizational values.

The result was An AI Roadmap for Nonprofit Arts and Culture Organizations, a practical guide filled with real-world case studies, lessons learned, and recommendations for responsible AI adoption in the nonprofit arts sector. You can download the Roadmap here.

Why A³ Matters

One of the most refreshing aspects of the initiative is what it isn't.

It isn't about chasing the latest AI tool. It isn't about replacing people. And it certainly isn't about asking ChatGPT to write your annual appeal and calling it a day.

Instead, the program encourages organizations to start with a business problem—not a technology.

Across the participating organizations, the strongest results came from identifying a specific bottleneck or repetitive workflow and then asking whether AI could help solve it. That might mean reducing time spent on grant research, improving internal reporting, streamlining marketing workflows, or creating better systems for audience communications. The emphasis remained on human judgment, organizational values, and measurable improvements.

Another theme that runs throughout the Roadmap is that successful AI adoption is as much about organizational change management as it is about technology. AI policies, staff education, and thoughtful experimentation are every bit as important as selecting the right tools. AI should expand human capacity and improve productivity of notoriously strained non-profit teams, not replace human expertise or decision-making.

As someone who has spent many years helping arts organizations grow audiences and strengthen their marketing, I found this philosophy especially compelling. Technology is most valuable when it improves workflows and eliminates repetitive, low-value tasks, giving people more time to focus on meaningful work and achieve results more quickly. Too often, I’ve seen talented teams stretched so thin that initiatives that should take a few months to complete end up taking years to come to fruition.

How Mark Morris Dance Group Put AI Into Practice

I was especially interested to see that Mark Morris Dance Group—where I previously served as Director of Marketing & Communications—was one of the sixteen participating organizations.

Rather than summarizing the published case study (which you can review on page 31 of the Roadmap along with the other participants’ case studies), I wanted to learn what the experience actually felt like from the inside.

I spoke with Lily McAteer, Institutional Giving Officer, who was part of the internal A³ task force, about why MMDG applied to the program, what surprised the team during the process, and how the organization thinks differently about AI today.

Laura Giannatempo: What prompted the Mark Morris Dance Group to apply for the A³ initiative? 

Lily McAteer: We applied to A³ because we wanted to understand how we could use AI to support our work as a nonprofit arts organization. We hoped it could help us work more efficiently and give staff more time for the relationship-based work at the heart of what we do. Essentially, we wanted to become better versed in the technology and to explore if integrating AI into our work could expand our capacity to serve artists, audiences, and community partners. 

We were also interested because the initiative felt like a natural continuation of the capacity-building work we’d done through Bloomberg Philanthropies' Digital Accelerator Program, which you led as Tech Fellow. Our experience in that program showed us that investing in new tools and ways of working could meaningfully benefit us as an organization and gave us confidence that participating in another tech-based capacity building program would be a positive experience. 

LG: Before participating, what was the staff's overall attitude toward AI? Has that changed since completing the program? 

LM: Both before and after A³, attitudes toward AI vary person to person. We have some early adopters who have been excited to explore this new technology, others who are more cautious, and some who are against its use. Across the board, though, I think everyone who is open to using AI in their work is interested in genuinely understanding best practices and how we can approach these tools thoughtfully. A³ definitely emphasized ethical, intentional use by looking at where AI can benefit what we do and where it’s maybe not necessary, which, in my experience, aligns with the staff’s intrinsic approach overall. 

The program gave us practical ways to evaluate those questions, like a questionnaire where we can assess what task an AI tool is assisting with, what the benefit will be (i.e. how much time will be saved, etc.), and if we feel this makes a solid case for using the tool. We also had an opportunity to become more skilled at prompting, so instead of running numerous prompts with questionable results, we could get what we needed out of one or two stronger queries. A³ reinforced the importance of always having human oversight: the so-called “human in the loop.” 

Organizationally, participating in the program led to the creation of an AI Committee (which I am on) and the development and adoption of a formal AI policy. The AI policy includes a request process for approved AI tools, which has been very popular with the staff. 

LG: Can you walk me through the experience? What did participation actually look like from your perspective? 

LM: We were selected from more than 200 applicants to join 16 peer organizations, including American Ballet Theatre, Alvin Ailey Dance Foundation, and the New Museum of Contemporary Art. Throughout the initiative, we were invited to participate in workshops covering topics like how to use AI ethically, effectively prompting, creating custom ChatGPTs, using Gumloop, and exploring existing AI integrations with tools like Tessitura. MMDG leadership and the AI committee attended all of the workshops, while staff across the organization were invited to opt into sessions based on their role and interests. 

We also participated in virtual meetups with staff from the other participating organizations; this was one of the highlights of the experience for me. I appreciated getting to hear how our peers were approaching AI, the questions and concerns they were working through, and how they were thinking about integrating these tools into their organizations. These conversations make the whole experience feel collaborative as we were navigating this rapidly evolving technological landscape together. 

LG: What are your key takeaways from your participation, besides your specific case study? 

LM: Our biggest takeaway was the necessity of establishing clear organizational guidelines for AI use, which led to the creation of our staff-wide AI policy which lays out clear expectations for the responsible, ethical, and secure use of AI tools. We also invited Katrina Ingram, Founder and CEO of Ethically Aligned AI, to lead a staff workshop on responsible AI use. She had led the first A³ workshop, and we found her insights so valuable that we wanted to share them with everyone in the organization. Katrina is the one who created the questionnaire I described in a previous answer, which has been very useful. 

On a personal level, I feel my prompting skills improved considerably, and I learned how to create custom ChatGPTs. They’ve become an important part of my work as the Institutional Giving Officer. Beyond the “Doc Weaver” template generator in our case study, I’ve found custom ChatGPTs and Claude projects can be helpful for discovering new foundation prospects, helping me time-block my day, and refining grant narratives to meet specific word or character limits. 

LG: Did anything about the program surprise you? 

LM: A³ was presented as an intensive “sprint,” and it absolutely lived up to that description. It covered a tremendous amount of ground in a relatively short period and could feel overwhelming at times. While I found the experience incredibly valuable and am grateful to have gotten to participate, it really was a chance to dip our toes into the world of AI. I left with as many, or more, questions than I came in with, which is probably healthy with a technology as new and vast as AI. 

LG: Where do you see AI fitting into your work at MMDG over the next few years? 

LM: In a big picture sense, I see AI fitting into my work the same way we all hope it will benefit the organization at large: supporting relationship-based work. How that looks from day to day may evolve, but that’s really the goal. 

I also think we’ll see more cross-departmental use of AI tools over the next few years. This might be through custom ChatGPTs, Claude projects, or workflow automation tools like Gumloop we create internally or through the continued AI integration within CRM platforms we already use; it will likely be a mixture of both. I see them becoming another part of how we work together across the organization.