Presented By O’Reilly and Intel Nervana
Put AI to work
September 17-18, 2017: Training
September 18-20, 2017: Tutorials & Conference
San Francisco, CA

Building an unbiased AI: End-to-end diversity and inclusion in AI development

Daniel Guillory (Autodesk), Matthew Scherer (Littler Mendelson, PC )
4:50pm–5:30pm Tuesday, September 19, 2017
Impact on business and society
Location: Franciscan AB Level: Intermediate
Secondary topics:  Law, ethics and governance (including AI safety), Organizational best practices
Average rating: *****
(5.00, 2 ratings)

Prerequisite Knowledge

  • Familiarity with the general concepts of diversity and machine learning

What you'll learn

  • Learn the different dimensions of diversity that are relevant to AI development

Description

Diversity has particular salience to designers and developers of AI systems. As AI systems perform more and more tasks that formerly were the exclusive domain of human professionals, it will be critical to design those systems in a manner that reflects the diversity of the societies in which they operate. Diversity in this sense certainly must start in the lab, but it cannot end there. The designers and developers of learning AI systems must also work to incorporate the interests and perspectives of the many stakeholders who could be affected by their systems’ operations and ensure that the datasets used to train their systems are representative of the relevant population.

Because all AI systems are programmed with default modes of operation, designers and developers must consider and integrate these human, cultural, and systems dimensions of diversity from the very beginning. If they fail to do so, they risk creating systems whose default mode makes them irrelevant to excluded groups. Worse yet, systems created without adequate attention to considerations of diversity and representation could make excluded groups themselves irrelevant or subject to discrimination.

Daniel Guillory and Matthew Scherer discuss the importance of ensuring diversity and inclusion when developing AI and share tips on how to do so. You’ll learn the urgency of proactively incorporating diversity and inclusion in the development of AI and leave with a framework for analyzing the various dimensions of diversity in AI development from both a legal and ethical perspective, along with practical tips on how to ensure diversity in thought, staffing, and datasets.

Photo of Daniel Guillory

Daniel Guillory

Autodesk

Daniel Guillory is the head of global diversity and inclusion at Autodesk, the leader in the future of making things, where he works to integrate all dimensions of diversity and inclusion into many parts of the organization, including customer acquisition, recruitment, hiring, people development, advancement, investment, and acquisition. He is interested in the application of people analytics to different initiatives. Previously, Daniel was CEO of Innovations International, a consulting firm that assists companies globally on leadership, innovation, and diversity through assessment, strategic planing, learning and development, and internal communications.

Photo of Matthew Scherer

Matthew Scherer

Littler Mendelson, PC

Matt Scherer is an attorney and legal scholar based in Portland, Oregon. He is an associate with Littler Mendelson, PC, and a member of the firm’s robotics, artificial intelligence, and automation practice group. Matt writes and speaks on the intersection of law and artificial intelligence and is the author of several articles, including “Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies,” which was published in the Harvard Journal of Law and Technology, and “AI in HR: Civil Rights Implications of Employers’ Use of Artificial Intelligence and Big Data,” which was published in the SciTech Lawyer. Matt also writes at Law and AI, a blog devoted to studying the emerging legal and policy issues surrounding artificial intelligence and autonomous machines.