This is a formal statement that Waymo knows people are going to get hurt in surprising new ways characteristic of AVs, and that they don't know how, but when they do get hurt it will be due to the "unreasonableness" of the situation, creating a carve-out for their defense during future litigation.
Your honor, as a prerequisite to review of the hazard of institutionalized overt pure design of mechanical harm, let's consider all the benefits that come from this technology.
Re-write the lede as "Safe for us to deploy, factors for plausible deniability and limits of liability."
Key points:
- We all did this together; read the literature, note all the stakeholders, consider our "aggregate", "holistic" approach to safety.
- There's "no user in charge", so no one to fault; it's all systemic edge cases.
- Look at our due diligence, we have ample data we submit as principled quantitative evidence that it hasn't gone too wrong.
- Legally speaking, we aim to dazzle with brilliance, but sometimes we have no choice but to baffle with BS.
- Philosophically, who's to say what's appropriate system behavior?
- To understand our approach, please refer to table 1 of 1 which is organized by the industry standard convention of rows and columns. In this table we set forth the language of criteria of all future litigation. Whatever goes wrong has a place in this table. To make litigation more approachable given the uncertainty of the venues in which we expect to defend, and the near total lack of a common vocabulary, we have helpfully pigeon-holed and enumerated the points around which we will organize every defense.
- Accountability, see figure 1 of 1, will be sloped and narrowed away from us and our release program and up towards the "safety board". They sign off and they're accountable, while we just build this stuff. When the Safety Board must answer to expected harms, a perfect cycle will be closed by the Safety Board referring down towards the Safety Framework Steering Committee. With concentrated denials and give and take, we'll minimize liability and reassure that whatever goes wrong won't happen again.
Finally, impugn us as you will, but such approach to liability mitigation follows an industry template and offers no innovations, so it can't be fair to single us out.
—Note list of citations showing that all this thinking came from elsewhere, so don't blame us.
fragmede · 9h ago
How we know Waymo are ready for the road? They've managed to get enough training data into their system to copy human drivers, which means they've started running red lights.
Your honor, as a prerequisite to review of the hazard of institutionalized overt pure design of mechanical harm, let's consider all the benefits that come from this technology.
Re-write the lede as "Safe for us to deploy, factors for plausible deniability and limits of liability."
Key points:
- We all did this together; read the literature, note all the stakeholders, consider our "aggregate", "holistic" approach to safety.
- There's "no user in charge", so no one to fault; it's all systemic edge cases.
- Look at our due diligence, we have ample data we submit as principled quantitative evidence that it hasn't gone too wrong.
- Legally speaking, we aim to dazzle with brilliance, but sometimes we have no choice but to baffle with BS.
- Philosophically, who's to say what's appropriate system behavior?
- To understand our approach, please refer to table 1 of 1 which is organized by the industry standard convention of rows and columns. In this table we set forth the language of criteria of all future litigation. Whatever goes wrong has a place in this table. To make litigation more approachable given the uncertainty of the venues in which we expect to defend, and the near total lack of a common vocabulary, we have helpfully pigeon-holed and enumerated the points around which we will organize every defense.
- Accountability, see figure 1 of 1, will be sloped and narrowed away from us and our release program and up towards the "safety board". They sign off and they're accountable, while we just build this stuff. When the Safety Board must answer to expected harms, a perfect cycle will be closed by the Safety Board referring down towards the Safety Framework Steering Committee. With concentrated denials and give and take, we'll minimize liability and reassure that whatever goes wrong won't happen again.
Finally, impugn us as you will, but such approach to liability mitigation follows an industry template and offers no innovations, so it can't be fair to single us out.
—Note list of citations showing that all this thinking came from elsewhere, so don't blame us.