Below is a summary of the key points I offered today at the Forum on Artificial Intelligence in Automated Mobility:
- Terms like “Trustworthy,” “Equity,” “Accountability,” “Ethics,” and “Human Centered” are frequently thrown around but usually without much attention to what they mean. Since these terms intrinsically sound good there is little to no effort given to providing a definition. This may in part be because the speakers themselves do not fully understand what they mean. This is not a benign omission and any claim that it is intentional is nonsense. My mission in compiling the AI Life Cycle Core Principles is to breathe meaningful substance to these principles.
- Clarity in the AI discourse is critical. It provides a common platform for advancing the conversation. Without it we cannot effectively calibrate and advance on the mission at hand. Clarity is a vital ingredient for meaningful regulations, laws, policies, and guidelines.
- There is significant emphasis on safety. But an Automated Driving Systems company that merely focuses on safety is missing the mark. Though important, an ethical-focused discussion needs to also consider other core principles, such as privacy and XAI.
- The fact that regulations oftentimes significantly lag behind technical innovation is not a bug. It is a positive attribute.
- It is necessary to resist the temptation to saturate technological developments with regulation. Succumbing to it is merely a knee-jerk reaction, a populistic attitude designed to satisfy critics and increase airtime.
- The six month moratorium on AI development that was called for earlier this year was nothing more than a PR stunt. Think about it! Why not five months? Why not eight? Of course, the outcome would have been that they need more time and there is little doubt no meaningful outcome would have been achieved anyway. Good thing it didn’t come to fruition.
- The existential risk theme that pervades some of the current discourse has its roots primarily in pop culture and science fiction. This feverish preoccupation distorts serious efforts aimed at increasing clarity on this subject. While AI is not poised to wipe out humanity, it certainly going to profoundly alter society. That’s not a bad thing.
- Just because something is explainable does not necessarily also equate with it being useful. Solely focusing on providing an explanation does not align with the Ethical core principle.
- Transparency that aligns with the Ethical core principle involves providing information that is accurate, sufficient, and useful. Deficiency in any of these three elements works to deteriorate alignment with the core principle.