New data and AI unit quietly absorbs five research units, costing survey research director his job

Over 28 years, Edward Freeland, former executive director of the Princeton Survey Research Center, helped hundreds of undergraduates, graduate students, and faculty develop survey research projects. When he was presented with a proposal in March to merge his center with four other University units focused on data, computing, and artificial intelligence research, he was excited to bring his ideas to the table.   

Freeland planned to test groundbreaking AI telephone interview services and systems for detecting bot survey respondents over the summer. Then, in May, he learned that not only was he excluded from the new academic unit, named Data and Intelligent Systems (DaIS) — he was being laid off. 

“Someone came from HR to show me the new organization chart,” Freeland said. “Two of my staff members were going to move over to it, but I was not in it.” 

DaIS — the result of a quiet summer merger of the Princeton Survey Research Center, the Center for Statistics and Machine Learning, the Data Driven Social Science Initiative, the Princeton Institute for Computational Science and Engineering (PICSciE), and the AI Lab — aims to “support Princeton’s artificial intelligence and data science research communities,” University spokesperson Jennifer Morrill wrote in a statement to The Daily Princetonian. 

Many of DaIS’s key operational details still remain unclear, even to faculty previously involved in the now-merged units, most of which had individual governing structures including faculty executive committees. DaIS currently has no online profile except for a lecturer job posting. In the last two faculty meetings, ownership of the undergraduate and graduate minors in Statistics and Machine Learning was transferred from the Center for Statistics and Machine Learning to DaIS.

“It isn’t clear to me, ​​as a faculty member, executive committee member, or affiliated faculty member, who is making the decisions, and when and why,” said English professor Meredith Martin, who was previously affiliated with or on the executive committee of four of the merged units. But she added that the lack of communication, both publicly and among faculty, is an effect of broad institutional change. 

“It’s not like [the merger] has been hidden. It’s more that the things are happening as quickly as they can manage, but institutional organizational work takes a lot of time, because when you change administrative lines, you have to change people’s jobs, and you have to do that really carefully, and create new positions, and move people from one to another in the HR landscape,” Martin said. 

Martin is also the founder and faculty director of the Center for Digital Humanities.

Over two dozen employees and affiliates of the units being merged into DaIS declined to comment or did not respond to requests for comment. The University provided a brief statement in lieu of responding to a detailed list of questions. It declined to confirm whether additional employees were laid off and whether the merger was a result of budget cuts, which led to layoffs in the Keller Center and the Trenton Arts Program in the spring. 

“This new unit has a nimble structure that allows Princeton to remain at the forefront of these fast-moving fields, while continuing to support student research and the minor and graduate certificate in Statistics and Machine Learning,” Morrill wrote. “DaIS will be a base from which the university can provide undergraduate and graduate populations with the coordinated training and skills they need to succeed when working with data and AI.”

DaIS, which will take over the Center for Statistics and Machine Learning’s home in Bendheim House, was created to promote collaboration between researchers and administrators previously separated both physically and intellectually, Martin said. The new unit will include two divisions, with one focused on statistics and data science and another on AI, she added. DaIS is the University’s latest addition to a series of AI-focused initiatives, such as the New Jersey AI Hub, which opened over a year ago.

“It’s really a way of actually optimizing administrative resources, and at the same time providing [a] conduit for collaboration across emerging fields that use the same technology, same technique for different goals,” said Luigi Martinelli, the new interim director of PICSciE and a mechanical and aerospace engineering professor. He said PICSciE became an initiative under DaIS — rather than an institute — because its research computing branch is being moved under the Office of the Dean for Research.

In his capacity as PICSciE interim director, Martinelli will serve on DaIS’s executive committee, which he said will handle the center’s strategic planning and maintain relations with other University units. The ‘Prince’ could not confirm whether the survey research center would have a seat on the executive committee. 

Morrill confirmed that DaIS will have two co-directors — politics professor Arthur Spirling, former director of the Center for Statistics and Machine Learning, and computer science professor Tom Griffiths, former director of the AI Lab. Remi Moss, currently listed online as the executive director of the AI Lab, will serve as DaIS’s executive director, Martin said. 

Freeland’s layoff has led to questions about whether DaIS will continue the Princeton Survey Research Center’s operations. Both sociology and public affairs professor Paul Starr and psychology professor Eldar Shafir, who are not involved in DaIS, predicted that the University’s survey research resources would take a hit without Freeland’s expertise and industry knowledge.

“The people who are moving from the [Princeton Survey Research Center] to [DaIS] might very well try to replicate all this. I’m not going to say it could never happen, but I’m very sure it’s not going to happen in the next year or two,” Shafir said. 

He argued that the layoff was handled abruptly and without regard for its downstream effects. 

“Those who made these decisions did not ask themselves the imputed cost of this move on students’ future research, on the seniors who are planning to do a survey this coming year, on faculty whose research had to stop in the middle,” Shafir said. “It was all done unceremoniously, with little fanfare, just announced and without any consultation, and that was very, very frustrating and not Princeton-like.”

Freeland helped ensure that survey research projects obtained real data, despite the proliferation of AI bot respondents that has called the validity of survey research into question, Shafir noted. Freeland estimated that he has saved the University millions of dollars either negotiating contracts with survey collection companies or administering survey projects himself.

“Freeland has provided not just technical advice, but also a great deal of counsel about how to do the research,” said Starr, who sent students to work with Freeland and personally co-authored an article with him. “Many more faculty have had the experience of doing large scale projects, part of which involved a survey, and he has been instrumental in making those things successful.”

Freeland sees his layoff as symptomatic of a broader trend in academia, where he believes survey research is losing ground as a priority. “In some ways, it’s viewed as old technology, so we’re either moving on to AI or other kinds of data science, trying to do more to extract data from these troves of administrative data or social media,” he said. 

In addition to questions about the prioritization of survey research, it remains unclear to what extent existing governance structures will be preserved alongside DaIS’s executive committee of directors. 

“My sense is that there isn’t consultation in the ways that I would have expected about governance, when there’s so many different faculty involved,” Martin said. 

She said that she was unsure what role she and other faculty previously involved in governance of the merged academic units will play in DaIS. But she explained that some of the now-incorporated units, such as the Data Driven Social Science Initiative, benefited from having executive committees of faculty with knowledge across a diverse range of disciplines.

“My sense is that the scenario would probably be that each of the places would have one faculty director who sits on … the bigger executive committee, but that wouldn’t be a logical body to make decisions about things like grants or research software engineers, because there’s a lot of domain specificity required there,” Martin said.

The merger will place each of the absorbed groups into its two divisions, statistics and data science or AI. PICSciE, which focuses on the “use of computation in simulation, modelling and analysis for scientific discovery and technology development” according to its website, may not directly map onto either of these divisions.

“Computational Science is a discipline per se which requires the numerical solution of complex  mathematical models of physical phenomena that are not necessarily data driven,” Martinelli explained in a written comment after the interview. 

However, he is confident that his field will remain a priority under the DaIS umbrella. 

“Being part of the executive committee of the entire operation guarantees that the voice and the priority of the computational science and engineering community are actually represented,” Martinelli said, emphasizing the importance of integrating AI and new numerical and algorithmic methodology into his field. 

“It’s going to get a huge multiplier, and I think that’s the rationale for doing this, but it is a work in progress, and we’re starting right now,” Martinelli said.

“We are excited about the potential of this new unit and look forward to sharing more with the university community as we move into the fall semester,” Morrill wrote. But that optimism about DaIS’ future was tempered, for some, by frustration over how it came about. 

“It’s funny to me because during the day we teach at Princeton about work relations and collegiality and respect and ethics, and then we act like a vulgar corporate entity, just firing people without much consultation,” Shafir said. 

Nico David-Fox is a head News editor for the ‘Prince.’ He is from Washington, D.C. and often covers breaking news. He can be reached at ndf[at]dailyprincetonian.com.

Please send any corrections to corrections[at]dailyprincetonian.com.