Soap Operas for Cars

The earliest prototypes of autonomous vehicles suffered a setback which plagued reinforcement learning for over 20 years, and even now leaves lingering doubts about the predictability of AI systems - the issue of recoverability. How do we teach a system to recover from a mistake without inadvertently giving it permission to make similar mistakes itself?
Just as the emergence of neutral networks came from our understanding of the brain, the solutions we've implemented reflect the human experience, from how we learn from failure, and the strategies we have at our disposal to prepare us for the unexpected.
Speaker Bio
Ike is a Senior Data Scientist at Elastacloud, with a background spanning Observability, Infrastructure Engineering, Cybersecurity and Fraud Detection. He originally started training as a data scientist in order to design better cocktails, and takes pride that he still has an edge over AI in this area.