Building the Future of Communications at Mayo Clinic: 7 Lessons Learned in Adopting Generative AI

The landscape of our profession is being transformed by generative artificial intelligence (AI) tools like ChatGPT, redefining what’s possible in communications. While some organizations have embraced these tools and are positioning themselves for the future, others have been slower to do so and risk falling behind.

Over the past year, the Mayo Clinic communications department has formally adopted generative AI and taken the first steps toward building a future with possibilities we’re just starting to imagine.

We’ve implemented dozens of use cases that save time and energy, generate insights, optimize content and elevate our work. Our media relations team developed a prompting guide to integrate generative AI into its workflows, and a growing team of super users is driving a culture of continuous learning.

Generative AI is not a passing fad. Effectively using it is becoming an essential part of the modern skill set for communicators, and taking the first steps in using these tools is imperative.

Here are key lessons learned in taking those critical first steps to generative AI adoption, based on our work at the Mayo Clinic communications department.

1. Leadership buy-in and support drive everything.

Dr. Halena Gazelka, Mayo Clinic’s chief communications officer, openly endorsed the adoption of generative AI at department meetings. The communications leadership team also empowered a small working group to move quickly with experimentation; benchmarking; drafting guidelines based on close consultation with legal, IT and information security; and developing an implementation strategy for department leadership approval. These steps enabled quick but careful progress with generative AI.

2. A tight timeline focuses effort.

In May 2023, we set an end-of-June 2023 deadline to finalize guidelines, hold key discussions with frontline supervisors and all department staff, and complete training. This ambitious yet achievable timeline accelerated progress while still allowing time for thoughtful discussion to inform our approach. These two months of activity brought the department onto the same page, established baseline knowledge and built the foundation for everything we’re doing today with generative AI.

3. Meet people where they are.

Throughout the adoption process, we created space for questions and concerns in leadership discussions, training sessions, one-on-ones and team meetings. These conversations helped build understanding and drove buy-in. Training included extensive demonstrations of practical use cases, such as drafting press releases and communications plans using ChatGPT. While we didn’t survey participants, anecdotally many shared that the hands-on approach helped answer questions and eased concerns about using these tools.

4. Follow existing policies — then be stricter still.

One of the biggest breakthroughs in creating guidelines was realizing that we weren’t starting from scratch. Mayo Clinic already has robust policies in place for protecting confidential business information, patient data, personally identifiable information and intellectual property. We also adhere to an organizational code of conduct, a model of professionalism and values that includes a focus on integrity.

With generative AI, our approach was to follow all existing Mayo Clinic policies — and then some. For instance, to safeguard against data privacy concerns, we disable data sharing in ChatGPT, employ generic language in our prompts and avoid using Mayo Clinic or any other specific identifiers. To address concerns about AI hallucinations, plagiarism, copyright infringement and bias, we emphasized the need for colleagues to (as always) be personally responsible for the integrity of their work and to safeguard Mayo Clinic’s reputation in their use of these tools.

5. Know your organization and be willing to lead.

Successful generative AI adoption requires honestly assessing your organization’s landscape. What's feasible given existing policies and where are the gaps? Mayo Clinic had a policy governing the use of AI in clinical settings, but nothing formally developed for non-clinical uses. We saw an opportunity for the communications team to lead other administrative areas in safe, responsible AI adoption. There was precedent for this leadership role, as veteran communications staff recalled the department taking similar initiatives with social media adoption 20 years prior.

6. Cultivate a community of learners.

To drive grassroots experimentation and deep generative AI adoption across the department, we formed a super user group from diverse teams. Membership has grown from seven individuals last summer to more than 30 today. This highly engaged group meets every three weeks to discuss the latest developments in generative AI, share learnings, take part in hands-on activities and inspire new use cases department wide.

7. This is a marathon, not a sprint.

The department has adopted what I consider a middle path strategy, balancing pace and caution. While initial steps toward formally incorporating generative AI require accelerated, focused effort, sustaining our evolution requires a consistent and sustainable approach. This balance is particularly crucial given the rapid advancements in generative AI. It's easy to be swept up in the excitement, but adhering to a structured process — like regular meetings with our super user team — provides a solid foundation for responsible and effective use of these transformative tools.

The Mayo Clinic communications department’s work with generative AI is just the beginning. With continuous learning, ongoing experimentation and a close eye on safety and responsibility, we are committed to continued growth and maturity in the use of these tools.

If your organization has taken the first steps to adopting generative AI, I encourage you to keep going. If not, I hope that the lessons shared here can be helpful in taking those first steps successfully.