Arguing Machines: Human Supervision of Black Box AI Systems That Make Life-Critical Decisions
This addresses safety and oversight issues in high-stakes AI applications like autonomous driving, offering a practical method for human-in-the-loop supervision without requiring access to system internals.
The paper tackles the problem of improving the reliability of black box AI systems in life-critical decisions by proposing an 'arguing machines' framework that uses disagreement between two independently trained AI systems to flag cases for human supervision, resulting in a reduction from 8.0% to 2.8% top-5 error on ImageNet and predicting 90.4% of challenging disengagements in Tesla Autopilot data.
We consider the paradigm of a black box AI system that makes life-critical decisions. We propose an "arguing machines" framework that pairs the primary AI system with a secondary one that is independently trained to perform the same task. We show that disagreement between the two systems, without any knowledge of underlying system design or operation, is sufficient to arbitrarily improve the accuracy of the overall decision pipeline given human supervision over disagreements. We demonstrate this system in two applications: (1) an illustrative example of image classification and (2) on large-scale real-world semi-autonomous driving data. For the first application, we apply this framework to image classification achieving a reduction from 8.0% to 2.8% top-5 error on ImageNet. For the second application, we apply this framework to Tesla Autopilot and demonstrate the ability to predict 90.4% of system disengagements that were labeled by human annotators as challenging and needing human supervision.