Uber's Approach to AI and Governance
Learn more about the latest AI initiatives at Uber
At Uber, we reimagine the way the world moves for the better. This relentless focus also applies to the responsible development and use of our technology.
AI is in our bones. Since our founding, we’ve used it to reshape transportation, enhance safety on the platform, and better connect people and cities. To do that, we’ve implemented governance that promotes the responsible development and deployment of AI technology every time we use it.
Uber’s AI principles
Uber employs AI throughout our platform to enhance daily experiences for our users. We developed the following 6 principles to guide us in the ways we use this technology to help them go anywhere and get anything.
Accountable governance
We maintain an AI governance program by documenting and operationalizing responsible AI policies, practices, and guidelines across Uber and seeking feedback for continual improvements to our program.
Responsible stewardship and trust
We establish and maintain a culture of ethical and human-centered stewardship of AI by designing systems for beneficial impact on people, the aggregate collection of communities we serve, and our business, laying the foundation for trustworthy AI as a shared responsibility across Uber.
Fairness and nondiscrimination
We incorporate fairness and inclusivity into our AI systems by designing, building, measuring, and monitoring them to treat all people fairly and address unfair bias or other unintended disparate impacts.
Transparency and understandability
We help make our AI systems understandable by designing them to be transparent about inputs and outcomes and communicating meaningful information to users.
Quality, safety, and security
We create robust, safe, and secure AI systems by evaluating system fitness, proper functioning, and impact throughout the entire life cycle.
Privacy and data protection
We respect privacy rights and protect personal data by evaluating the quality and use of such data in our AI systems. We also follow our published Privacy Principles
Uber’s process for AI accountability
ONeil Risk Consulting & Algorithmic Auditing (ORCAA) is an organization dedicated to defining accountability in algorithms and keeping people safe from these systems. The organization is led by Cathy O’ Neil, author of Weapons of Math Destruction.
"Overall, ORCAA finds that the governance structure is successful in achieving two critical things: first, a broad definition of AI risk, which is to say one that accounts for external as well as internal stakeholders, and second, a clear process for composing and maintaining accountability for that risk."
Team spotlight
Sean Perryman | Head of AI Fairness Policy
“AI is the backbone of the Uber marketplace, but we realize that our users often don’t think about it when visiting our platform. They want to get where they need to go or receive something they may need for their daily life. We want to make sure our technology works seamlessly and is safe and reliable. When we consider our approach to governance, we want the technology to work for users in a way that they expect and to make sure they feel like they’re not being treated differently based on who they are.”
Ian Kelley | Head of Strategic Initiatives
"Machine Learning is at the heart of many of our company’s core customer experiences, from predicting how long it will take for someone to get somewhere in an Uber to providing personalized recommendations in Uber Eats. As the power of AI has expanded, so too has Uber’s need to ensure we use it responsibly, allowing us to create amazing experiences while protecting our customers’ trust in our company and its products."
Dan Svirsky | Applied Economist
“We have the words for what we want in AI governance: fairness, equity, explainability. But how do we make the translation from idea to practice? It’s a challenge for everyone, and we here at Uber are excited to do the work.”
Melissa Barr | Senior ML Technical Program Manager
“AI is a foundational pillar of the Uber platform. It’s ingrained in each step of the user experience, but often behind the scenes. That’s why we’re committed to embedding accountable and transparent practices into our AI systems, through building trustworthy tools that not only serve our communities but enhance safety, security, and understanding. This leads us to a future where technology operates with integrity, respects privacy, and promotes fairness for all.”
Our approach
The Legal, Tech, and Policy teams at Uber have established a variety of groups and technologies to support how we manage AI strategies and products:
AI Law & Ethics Council
This cross-functional group of leaders is responsible for overseeing AI governance and strategies, plus collaboration on policies and guidelines needed to operationalize AI governance at Uber.
Fairness Working Group
This cross-functional group of employees across the Legal, Policy, and Engineering teams identifies products for testing and mitigation of technical or algorithmic issues.
Marketplace Fairness
On this team, data scientists test products, processes, and algorithms for inadvertent disparate impact.
Michelangelo
Uber’s Michelangelo is a machine learning platform that centralizes model information and allows for monitoring of versions and outcomes.
Office of Responsible AI
Sponsored and co-chaired by Legal, this team implements OKRs (objectives and key results) approved by the AI Law & Ethics Council. That includes driving improvements in model assessments and helping to integrate a culture of responsible AI throughout the company.
Product Equity
Another cross-functional team, this one focuses on creating accessible, inclusive user experiences that favorably address the perspectives and needs of historically and systemically excluded communities.
How Uber’s AI initiatives make a difference in the real world
Automations
Use support bots and document automation to streamline operations.
Efficient marketplace
Improve matching, routing, dispatching, pricing, and incentives, and enable features such as Upfront Fares.
User experience
Tailor experiences in Uber Eats, search, and rides based on your prior usage and marketing preferences. Enhance the experience for consumers, drivers, and couriers with personalized suggestions and support.
Productivity
Provide AI-driven tools for coding, data management, and automated testing and fixes.
Risk and safety
Support ID verification, fraud detection, and incident prevention.
Ongoing AI innovation
With AI and machine learning at its core, Uber is on an ambitious trajectory to enhance our offerings with cutting-edge generative AI technologies. Because we view AI governance as a journey rather than a destination, we’ll continue to evaluate and implement them over time as best practices change.
Internally, we’re empowering our technical teams with these advanced tools to drive productivity and innovation. By embracing such transformative technologies with a safety-first mindset, Uber is committed to setting new standards in our industry, continually enhancing the user experience, and achieving operational excellence.
Earn
About
Earn
About