In 2017, artificially intelligent (AI) technologies surged into the popular discourse for its advancements — such as autonomous vehicles and predictive analytics — to critiques of potential biases, inequity and need for transparency. Growth in dataset sizes, increased computing efficiency and enhanced techniques in neural networks and machine learning are all factors in the success and pervasiveness of AI systems. These very same factors, however, also contribute to its limitations and biggest challenges. The prevailing unknown remains: to what extent do AI technologies affect our society’s basic institutions, daily practices and cultural norms?

For the second-year, the Aspen Institute Communications and Society Program convened thirty thought leaders from across disciplines and various fields to question the many ways in which AI may or may not impact the world. Participants of the Roundtable on Artificial Intelligence, held in August 2017, challenged the powerful narrative of AI’s growing dominance and inevitable influence on today’s society. To facilitate this thinking, the Roundtable encouraged participants to re-situate the human as the focal point, asking the question: In what ways do AI innovations enhance and or limit personal human autonomy? Discussions traversed the philosophical to the applied, and raised deep, fundamental questions on how to guide the trajectory of AI “with meaning and dignity for humans.”

The following report, “Artificial Intelligence, The Great Disruptor: Coming to Terms with AI-Driven Markets, Governance and Life,” authored by David Bollier, brings to light various nuanced issues in assessing tradeoffs between the benefits of AI and the potential negative social consequences.

The report is divided into four sections, which reflect the key themes by roundtable participants addressed. First, “The Far-Reaching Disruptions of AI,” highlights key technological trends across nine industry verticals and calls into question the techno-deterministic narrative so widely ascribed to AI systems. Second, “The Co-Evolution of AI and Humanity,” reflects on four historical critiques of technology and begins to unpack the human and machine relationship.

In the third section, “The Perils of Predictive Analytics in Criminal Justice,” Bollier provides a concrete example of the complications embedded in AI systems, such as automatic sentencing programs and predictive policing. As several participants voiced, a move towards eliminating human discretion to an automated justice fails to account for structural inequalities and obfuscates accountability. The section also cites several citizen-data projects that aim to neutralize biases in data and algorithms.

Finally, in the fourth section, “AI Governance and Its Future,” the report outlines the shortcomings of AI governance models in both the U.S. and around the world, leaving open questions on how best to evaluate AI innovations according to pre-existing standards, and under whose authority. Recommendations on developing a “soft law” versus strict regulation may help keep pace with the speed of change in AI. Most notably, the discussion on AI governance by design offers a starting framework to assure accountability for AI that maps onto the very basic questions of who, what and when.

At the end, the conference confirmed that the impact of AI technologies will be measured not only by units of its adoption but by the breadth of human success and failure. We hope the report provides the humanistic lens through which industry, government, academia, civil society and citizens can better address the future of the interaction between human and machine.

Acknowledgments
On behalf of the Aspen Institute Communications and Society program, I want to thank Michael Ferro and the Ferro Institute for their continued support and leadership in developing this roundtable. Thanks, also, to David Bollier, our rapporteur, for capturing the various dialogues, debates and nuanced viewpoints of participants. As typical of our roundtables, this report is the rapporteur’s distillation of the dialogue. It does not necessarily reflect the opinion of each participant in the meeting. Finally, I want to thank Dr. Kristine Gloria, Project Manager, and Tricia Kelly, Managing Director, for their work on the conference and bringing this report to fruition.

Charlie Firestone
Executive Director
Communications and Society Program
The Aspen Institute
February 2018