News & Events

Upcoming Events

Start Date/Time:
15/10/2018 7:00 pm
End Date/Time:
15/10/2018 9:00 pm

Sopra Steria
120 Holborn

We are right to make every effort towards safe AI, which is aligned to human values, but if we want it to happen at all we must have some tolerance for risk.

Autonomous vehicles lay the issue out clearly. Today, they are not safe enough, with the best systems needing human intervention every 5,555 miles, compared to 165,000 miles average between human-caused accidents. They are not ready in their present state, but if we have a complete zero tolerance for mistakes they’ll never be ready.

The focus of this meetup will be demonstrating why some level of risk is unavoidable and how we can reach a point of consensus on the acceptable level of risk where AI development and implementation can continue.