KEsu Origins February 24 — 26, 2017 PROJECT An Origins Project Scientific Workshop ARIZONA STATE UNIVERSITY Challenges of Artificial Intelligence: Envisioning and Addressing Adverse Outcomes In early prototypes, information was provided visually to the healthcare provider. However, it was quickly discovered that if audible feedback was provided — as if the Al device was a consultant working with the healthcare provider and the patient — the consultations were more efficient; patient satisfaction levels were higher; and outcomes were more positive. A large segment of health insurance sector sees early wins in supporting the technology, through the ability to decrease insurance claims through efficient and preventative interventions, while maintaining high premiums. As such, they push for early and widespread adoption of the technology. This is further supported by the Department of Health and Human Services as it hits a number of goals, including increasing health and well-being while reducing healthcare costs. With the success of early implementations, new Al-based technologies are rapidly implemented into subsequent generations of Al-consult. However, the commercial sector developing and using Al- consult has shifted dramatically from the technology’s initial beginnings. As the technology began to mature and lead to substantial savings in healthcare costs traditional healthcare providers and health insurance companies begin to suffer. They resist the use of Al-consult through a combination of lobbying for new policies and regulations limiting use, to marketing campaigns persuading people of the critical importance of human interaction in healthcare. They forge links with a number of advocacy groups opposed to widespread automation in society, and promote the idea of Al-consult undermining human dignity and jobs creation. However, the health benefits and cost savings of Al-consult are so compelling that these campaigns gain little traction. As a result, companies th