Oct 8, 2019
Federal CIOs Talk AI and Federal Service at Harvard
What happens when you put Harvard College, Engineering, Business and Kennedy School of Government students together with the federal government CIOs and talk artificial intelligence and automation?
HBS Professor Mitch Weiss and I thought we’d find out.
With students from the HKS Data Science Club, the Harvard Open Data Project, the HBS Government and Public Policy Club, and Coding It Forward in attendance, the audience was curious to hear how federal CIOs are applying artificial intelligence and machine learning (AI/ML) and automation in government IT.
The short answer from Suzette Kent, US CIO, is that AI/ML is already being used across a number of government mission areas. As one example, she cited a disaster recovery algorithm, developed in a national research lab, that takes data from multiple agencies to suggest the optimal placement of ambulances for natural disaster responses.
The rest of the panelists gave additional examples:
- Gary Washington, USDA CIO, highlighted the use of AI/ML and automation in food safety inspections;
- GSA Deputy CIO Beth Killoran discussed how AI/ML is improving procurement and real estate at GSA; and
- DOJ CIO Joe Klimavicz and FBI IT executive Rich Haley discussed how crime scene processing and cybersecurity investigations rely on modern data processing and AI/ML.
Will humans be removed from the loop, Mitch asked?
All the panelists said that for the foreseeable future, humans will still be in the loop making critical decisions — AI will augment, automate, and support the work of federal officials. Unlike rules-based software, AI will present new challenges on oversight and explainability, the CIOs noted.
The panelists said that government also has an important role to protect sensitive data, as well as share data internally and externally responsibly. Suzette Kent highlighted the federal data strategy action plan and recent legislation that calls for the creation of Chief Data Officers and infrastructure and processes to make data more usable.
One question which emerged: should the government be able to share data internally to better anticipate service delivery and make better decisions — even if the government is using data beyond its originally intended purposes? Several students in the room generally leaned forward and wanted a more efficient and coordinated government; the CIOs agreed in principle but also noted the practical legal restrictions on the use of certain data in government.
The students wanted to know how the CIOs approached change management, given the rapid pace of change in technology. Mitch teaches a popular HBS course on public entrepreneurship; I teach a field class where students work closely with a government client to have an outsized impact.
We too wanted to know how the CIOs thought about change.
While modern technology is critically important, it’s not always enough, the CIOs explained. Suzette Kent noted: “Sometimes the technology is the easy part. Getting the rest of the organization to change is the hard part.”
Students had asked what they should be learning to prepare themselves for public service, and the discussion turned to being able to focus on user needs (e.g. design and product management skills), and how government needs people to translate between data scientists and the bureaucracy.
In the cyber area, there are many thousands of open positions in federal government, the CIOs noted. In addition to hiring new employees, the federal government is working to reskill employees with initiatives like the Cyber Reskilling Academy.
The session ended with a call to service by the CIOs, as they talked about a scale, scope, and mission of federal government that is hard to match. And with new direct hire authority for CIOs to hire technical talent, they emphasized, and programs like USDS, 18F, and Presidential Innovation Fellows that hire for shorter terms — there are more ways for technical, design, and product-oriented students to serve in federal government.
