One of the genuine pleasures of serving as ASA president is engaging with dedicated colleagues on questions that matter deeply to the profession. Every board meeting offers those moments, but the spring meeting stands out. In April, we had an extremely productive strategic dialogue about artificial intelligence and what the ASA is going to do.
We know the conversations occurring in policy, industry, and government about how to develop and deploy AI responsibly are happening, far too often, without statisticians in the room. This is all set within a broader landscape in which a statistics department at a major research university closed and our early- and mid-career members are expressing genuine anxiety about what AI means for their work and careers. These are not isolated signals, and they demand a response.
The ASA has been paying attention. We have a growing body of member engagement on the topic. But statements and attention are not the same as strategy. This spring, the ASA Board committed to moving from awareness to action.
Ethan Mollick, a professor at Wharton, writes a Substack column called One Useful Thing. He ends a post titled “Choosing to Stay Human” with this advice:
More broadly, we are at the point where the defaults are being set for what kind of work to give AI: by the AI companies designing for frictionless use, by employers deciding what counts as “using AI well,” and by people teaching the ever-shifting concept of “AI literacy.” A lot of this is happening without, ironically, any real planning or consideration. And I suspect it will be hard to reverse these defaults once a generation of workers and students has built habits around them. The most important thing we can do is keep asking what to hand over and what to keep for ourselves … and not expect anyone, including the AI, to answer that for us.
Our strategic plan will help guide what we “hand over” and what we “keep for ourselves.”
How We Worked
The centerpiece of that commitment was a two-hour structured strategic planning session at our spring meeting, co-facilitated by ASA Executive Director Ron Wasserstein and me. We did not begin with a blank page. In the weeks before the session, staff assembled a curated reading repository drawing on the ASA Statement on the Role of Statistics in Data Science and Artificial Intelligence, “Rebuilding Statistics in the Age of AI: A Town Hall Discussion on Culture, Infrastructure, and Training,” and the AI Index, an initiative at the Stanford HAI. Board members were asked to come prepared, and they did.
We worked in two rounds of small groups, each focused on one of the following five strategic domains that emerged from the ASA’s community discussions and the 2024 JSM town hall on rebuilding statistics in the age of AI:
- Culture and Practices of Statistics
- Data Work and Infrastructure
- Engaging with AI/ML Innovations
- Next-Generation Talent and Education
- Stakeholder Engagement and Advocacy
Each small group identified opportunities, named risks, and articulated three-year visions. Then the full board voted to identify our three highest priorities. During a second round of breakout sessions, groups drafted concrete SMART objectives and action steps for each priority.
What emerged was not a finished plan but something more valuable at this stage: a shared framework, a set of real commitments, and a draft scaffold that—with your help—will be developed into a full strategic plan for ratification at our JSM board meeting.
The Three Strategic Priorities and Draft Objectives
After rich discussion, the board determined the five domains could be meaningfully addressed through three strategic priorities, with two—Data Work and Infrastructure and Engaging with AI/ML Innovations—neither abandoned nor overlooked, but deliberately incorporated into the three priority domains. I want to share the draft objectives here not as finished language but as evidence of the seriousness of the work and as an invitation for the community to engage with what we are building.
The first priority is Stakeholder Engagement and Advocacy. One of the clearest conclusions from our session was the statistics profession’s unique contributions to responsible AI development are not understood by the broader public, policymakers, or even many of the technologists building these systems. That is both a failure of communication and an advocacy opportunity.
Our draft objective reflects this: By December 31, 2026, the ASA will identify priority stakeholders across policy, industry, funding, and the scientific community and launch targeted engagement and advocacy campaigns, anchored in the message that AI must not be developed or deployed responsibly without statistical expertise, as measured by the number of stakeholder engagements, media placements and impressions, formal partnerships or coalitions established, and ASA presence in AI governance and standards conversations.
The second priority is Next-Generation Talent and Education. Statistics training has not kept pace with where our graduates are going or what the field needs. For example, we should teach machine learning through a statistical lens asking when performance generalizes, how to quantify uncertainty, and what assumptions justify a prediction. Teaching ML purely as a set of computational techniques, without this lens, surrenders everything that makes statistical training distinctive.
Our draft objective is: By JSM 2027, the ASA will develop and disseminate a white paper on statistics curriculum for the AI era, offering concrete recommendations for preparing undergraduate and graduate students for leadership roles in the ethical advancement and use of AI across academe, industry, and government and will launch an accompanying outreach effort to encourage curriculum modifications consistent with its recommendations, as measured by dissemination reach, the number of departments reporting curriculum modifications consistent with the recommendations, and integration into ASA education and professional development programs.
The third priority is Culture and Practices of Statistics. This is the hardest domain and the one in which we were most deliberate about not settling for a narrow answer. Professional development matters. Statisticians need new skills and fluency to engage confidently at the AI frontier. But culture is deeper than skills. It is about what we signal we value—in our publications, in our awards, in the language we use to describe excellent work, in how we support early-career members who are navigating genuine uncertainty.
Our draft objective is: By JSM 2027, the ASA will take deliberate steps to shape the culture and practice of the statistics profession in the AI era, including how the ASA recognizes valuable contributions, communicates the profession’s identity, and supports members’ confidence and sense of purpose, while expanding professional development offerings developed in collaboration with statistical leaders in AI, as measured by the breadth and participation in new professional development offerings; member surveys on professional confidence and identity; and observable changes in ASA publications, awards, and public communications that reflect the profession’s evolving identity in the AI era.
A draft strategic plan reflecting this work is now available for community comment through July 15. The board will ratify the final plan at the Joint Statistical Meetings in August.
I want to share something that was said during our session’s opening framing and has stayed with me: Whether statistics emerges from this period diminished or renewed will depend on how deliberately it adapts. The board’s work this spring was an act of deliberate adaptation. The plan we are building is a commitment to see it through.
The community’s voice will be essential in shaping what comes next. I look forward to your comments and the conversations we’ll be able to have at JSM. As you plan your schedule, please add Peggy Carr’s talk, “From Lived Experience to Public Trust: Stewardship, Service, and the Future of Federal Education Data,” to your must-see list. I’m grateful Peggy accepted my invitation to give the President’s Invited Address on Monday, August 3, at 4 p.m. JSM is truly a celebration of our community’s accomplishments, and I look forward to congratulating the award winners in person after my talk on Tuesday before we all unwind at the Dance Party.
i think one of the biggest problems is that as statisticians we never fought for ourselves in the first place and now with AI around, we are fighting for our survival. This is exactly now what we will have to do. I think one of the biggest elements we are ignoring is not just integrating ML into our curriculum but also preserving many elements of our statistical methods and why they still need to taught and will also be useful in the ML framework. I never see that being proposed. I just keep seeing that we have to catch up and revamp and that we will lose out if we do not. We also need to give some pushback to the total inclusion into AI, which I hate to say that some of our leaders advocate for instead of also advocating for our profession. Also, real, succinct guidance needs to be developed from our side. This is all what I am hoping to see and would also like to promote or we will be gone. Please let’s continue fighting for our profession and integrate gracefully.
I feel that in a data science program curriculum, the ideas that are developed (maybe parallelly) in statistics, and computer science need to be connected and compared. Many ideas were borrowed and used from statistics but labeled with a different terminology system. Given more information about this would enhance our profession through the cognition of the data science practioners.
Statistics has never been just pure mathematics and algorithms.
It uses mathematics to evaluate factual situations.
Every mathematical method requires prerequisites, under which it may be applied and under which it can also yield meaningful results.
A key task of the statistician is to verify that these conditions are met, or at least to assess the plausibility of their fulfillment.
They document the results of the assessment, any assumptions made, and the rationale for their plausibility.
They must explicitly point out the consequences if any assumptions made are not valid.
That does not change in the age of generative AIs. In fact it becomes more important than ever.
The uses of data that people care the most about involve prediction in one form or another; this was the primary point I was trying to make in the article “The Need for More Emphasis on Prediction: a ‘Nondenominational’ Model-Based Approach,” which was published (with discussion) in The American Statistician in May 2014 (Volume 68, pages 71–92). That point is not reflected in the traditional teaching and practice of statistics. When it comes to prediction, the public is much more inclined to think of AI than of the things they associate with statistics, and the statistics community has mostly itself to blame for that. If the statistics community hopes to fend off the inroads and threats posed by the advent of AI, it needs to make major changes in the teaching and practice of statistics, so as to much better reflect the uses of data in which people are most interested and most care about.