Soon after profitable awards for specialized advances in financial idea and industrial group, serving as main economist at Microsoft, conducting first investigation that combines device finding out with econometric modeling, and groundbreaking the new subject of tech economics, Susan Athey will now check out on a new hat as main economist at the antitrust division of the U.S. Division of Justice.

The broad-ranging mother nature of her work and vocation has garnered her regard throughout numerous tutorial fields. “Susan is a pressure of nature. She moves from machine understanding to company strategy to technological know-how policy to social effects, manufacturing deep concepts at each individual transform,” says Jonathan Levin, the dean at Stanford Graduate College of Business enterprise, exactly where Athey is the endowed Economics of Engineering Professor.

Even though Athey will carry on her Stanford GSB appointment in a portion-time capacity as she ways into her new govt part, the change in her target presents an option to reflect on the significant effects she has experienced all through her job and whilst at Stanford.

Introducing AI to Economics

Athey’s propensity for immersing herself in diverse fields of analyze goes back again to her undergraduate days at Duke University, where she graduated in 1991 with a triple-main in economics, arithmetic, and laptop or computer science.

Just after earning her PhD at Stanford GSB, she went on to grow to be a professor of economics and organization at MIT, Harvard College, and then at Stanford starting off in 2013. Even within the area of economics, Athey’s passions have been diverse: In 2007, she received the prestigious John Bates Clark Medal for her contributions to many subfields, like industrial firm, microeconomic concept, and econometrics.

But it was through a depart from academia to provide as Microsoft’s chief economist from 2008 to 2013 that Athey produced a stunning connection concerning her enthusiasm for economics and the applications of AI and equipment learning.

Athey already knew that digitization and tech platforms ended up likely to play a sizeable role in the economic system, and that search engines had been poised to have outsized value. She also understood that the investigation community was just beginning to deal with concerns about how to structure electronic marketplaces and what balanced competition appeared like in individuals marketplaces, and she was energized to aid establish that research.


Because technological innovation these types of as artificial intelligence moves so speedily, it is tough for the government to maintain up. We have to figure out how all branches of federal government are likely to be ready to guidebook us through a different age.

But the moment she started functioning at Microsoft, Athey also identified anything she was not expecting: the possible for device discovering to deal with financial challenges. The creators of the Bing research motor were being conducting experiments in a way that economists only dreamed about. They ended up simultaneously managing thousands of randomized A/B assessments — inquiring massive figures of “what if” queries to better recognize this kind of matters as which search benefits ought to rise to the leading and how to operate auctions for setting promoting prices on a research website page. By comparison, she suggests, economists would normally run a single experiment in a 12 months.

“Microsoft was making use of an artificial intelligence method composed of hundreds of algorithms that had been all operating alongside one another to make a look for final results webpage,” she suggests. “That was some thing new.”

Machine Discovering and Causal Consequences

In the subject of economics till then, data mining and equipment studying experienced been pejorative conditions for a considerably less innovative variety of statistics. “They were observed as a mechanical workout to different cats from canine,” she claims. But at Microsoft, Athey saw an option to blend the computational advances from predictive equipment mastering with statistical principle so that scientists could better comprehend causal outcomes not only in business enterprise programs like the research motor, but also in social science and economics. It was an epiphany that introduced her in a new investigate course and helped define her as one of the early tech economists.

Coming out of her working experience at Microsoft, Athey recognized that the insights from predictive algorithms could be harnessed in new techniques by combining them with the latest developments in econometrics and figures. For example, equipment discovering algorithms could be tailor-made to respond to induce-and-impact questions in economics, this kind of as what will occur if we change the least wage? Increase immigration plan? Increase price ranges? Make it possible for two companies to merge? “Predictive equipment discovering can’t resolve these questions alone, but it can help,” she suggests.

For illustration, Athey has utilized machine discovering to seem at the effects on buyers of customized pricing, a type of cost discrimination that involves charging unique charges to customers in accordance to their willingness to fork out. Regular economics approaches would give combination methods to that difficulty, she claims. They would examine maybe one product class at a time, contemplating demand for, say, various brands of yogurt or towels. By making use of device understanding strategies to consumers’ historical invest in facts, Athey’s research group can estimate personalised customer tastes across various solutions at the exact same time.

Creating these predictive styles of consumer decision in switch permits researchers to talk to even larger concerns about this sort of issues as what transpires to selling prices if you apply a tariff, or if generics arrive on the sector. “As an input to answering these inquiries, we want to recognize how consumers make selections,” Athey says. And machine discovering provides that enter in a way that will allow researchers to do this operate additional efficiently, at a more substantial scale, and in a a lot more individualized way. “If you are assuming everybody’s the exact same, that presents distinctive responses than if you think that individuals have different preferences,” she suggests.

A Tech Economics Pioneer

Athey’s main economist role at Microsoft finished in 2013, but her tenure there defined her as one of the very first individuals to be deemed a “tech economist.” It is a discipline she has considering that aided establish as an impartial self-control by convening early conferences in the industry and mentoring several college students together that occupation route.

“Now tech economists hold an annual convention that draws about 800 individuals,” she suggests. “And we have a specialised career marketplace mainly because getting a tech economist is a distinct career that men and women can go after.”

Athey has also created about what it means to be a tech economist. “It’s partly a occupation, but it’s also a mix of distinct fields of research,” she says. Tech economists research the influence of digitization on the financial system, which involves wondering about industry structure, privateness, details safety, fairness, competition policy, and more, she claims. “They also support develop and review small business types and competitive tactic, and they link the types to data to manual conclusions.”

At Microsoft, in addition to having an sudden deep dive into device studying and AI, Athey witnessed firsthand the troubles created by these systems — moral and lawful troubles, To start with Amendment difficulties, fairness and bias, privateness and copyright, and the prevalence of unexpected outcomes as people today manipulated or gamed the technique in response to market place shifts or new regulations.

Due to the fact of these observations, Athey created a drive to influence the approaches that device discovering and AI would play out in the environment. When she returned to academia total time, her initial techniques in that course included encouraging to prepare the launch of HAI and then starting to be a single of HAI’s founding associate administrators. “Stanford HAI was genuinely designed to handle these complications,” she suggests. “We want to make AI helpful for individuals, and we want to avoid all of these unintended consequences.”

Athey also required to translate the profitable works by using of AI from the for-revenue sector into the social impact sector. This urge led her to start the Golub Capital Social Affect Lab. “We’re bringing the tech toolkit to social influence purposes,” she states. So, for case in point, the Social Impression Lab has performed case studies of digital education technological know-how to enhance students’ discovering, created and executed approaches to focusing on instructional messaging to increase engagement of farmers, made and evaluated electronic tablet applications that guide nurses by counseling clients, and designed approaches to prioritize candidates for clinical trials of COVID-19 prescription drugs.

Connecting the Dots to the DOJ

Applying equipment learning to interesting social troubles at the Golub Money Social Effect Lab is a bottoms-up solution to earning change, Athey claims. By contrast, in her new task as chief economist of the antitrust division at the DOJ, she will be seeking her hand at addressing the issues of the digital financial system from the major down. “Government rules and policies impact every little thing from how competition functions to what mergers go as a result of, to what investments folks make,” Athey says.

Athey is also a founder and associate director of the Stanford Institute for Human-Centered AI. In relocating to the DOJ, Athey hopes to carry on many of Stanford HAI’s efforts to help governments adapt to an period of quickly altering technology, significantly all around the use of details in business and in governing administration. “Because technologies these as artificial intelligence moves so immediately, it is hard for the authorities to continue to keep up,” she claims. “We have to figure out how all branches of governing administration are heading to be well prepared to information us via a diverse age.”

It’s a good time for Athey to consider her hand at authorities do the job, says Levin. “At a instant when technology is ascendant and selling level of competition is crucial, I simply cannot think of everyone I’d instead have at the DOJ than Susan.”