Technology & Data Focus Areas
Explore the technology and data issues that inform our research, learning, partnerships, and advisory work—and our perspective on how they should be governed and applied.
Artificial Intelligence and Responsible Automation
What it is
Artificial intelligence and automated decision systems increasingly influence hiring, benefits, lending, health care, education, and public services. This area covers the models themselves, the data that trains them, the institutions that deploy them, and the oversight arrangements that determine whether people can understand, question, or appeal a decision made about them.
“An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.”What this looks like in everyday life
A short story you can click through, one step at a time.
Amina applies for help paying for school. A computer reads her form first.
The computer learns
The computer studies thousands of old applications to spot patterns. If those old files were unfair, it learns the unfairness too.
The point: Automation is fine when a person is still answerable and you can challenge the result.
Why it matters
Three reasons at a glance — hover or tap a card to read more, and follow the link to the source.
People
Scored, sorted, and ranked
Source: NIST AI Risk Management Framework 1.0Automated systems now sit inside hiring, lending, benefits, and clinical triage, so ordinary decisions about people are increasingly mediated by models rather than by a person they can talk to.
Fairness
Error rates are never even
Source: Buolamwini & Gebru, Gender Shades (PMLR, 2018)Audits of deployed systems repeatedly find accuracy that varies sharply by skin tone, gender, accent, or income, meaning the cost of a mistake falls hardest on the groups already least able to contest it.
Recourse
Hard to question, harder to fix
Source: Ada Lovelace Institute, Algorithmic Accountability for the Public SectorWhen a model's reasoning is undocumented, affected people cannot see why they were refused, and institutions cannot detect drift or harm until it has already scaled across thousands of decisions.
Our perspective
DPA supports the use of AI where it demonstrably improves outcomes, and expects the burden of proof to sit with the institution deploying it. Responsible practice means documenting purpose and limitations, testing for unequal error rates before deployment, keeping a person accountable for consequential decisions, and giving affected people a plain-language explanation and a usable route to challenge outcomes.
Connected pillars
- Research & Insights
- Talent Pathways & Readiness
- Learning & Convenings
- Advisory
Showing Artificial Intelligence and Responsible Automation
Select a Focus Area
Technology and data issues rarely exist in isolation.
DPA works across disciplines and sectors to understand how these systems intersect—and to help communities and institutions navigate them responsibly.