Here's a summary of the provided text, including all the key assertions made:
The speaker argues that many people haven't grasped the fundamental distinction of Tesla's Full Self-Driving (FSD) system. They highlight several points about Tesla FSD:
1. **Vision-Only & Generalized Platform:** Tesla FSD is a vision-only, generalized autonomous platform that perceives the world through cameras, much like human eyes.
2. **Training Data:** It's trained on video of "good human drivers" encountering various situations (e.g., flooded streets).
3. **Massive Scale Training:** This training leverages an "enormous amount of high quality data" at "massive scale" due to Tesla's existing fleet of "roughly 10 million vehicles," resulting in "thousands if not tens of thousands of training examples."
4. **Human-like Adaptability:** Tesla's vehicles are claimed to "know how to drive period," similar to how humans can drive in new cities, states, or even countries (including those with opposite side-of-the-road driving) with minor adjustments.
The speaker then criticizes "so-called experts" for an "own goal":
5. **Experts' Flawed Belief:** These experts allegedly believe that safe autonomous software *requires* "every sensor on earth" (Lidar, Radar, etc.).
This belief is then directly contrasted with Tesla's claimed performance and Waymo's alleged incidents:
6. **Tesla FSD's Performance:** Tesla vehicles are reportedly "operating fully autonomously now in almost half a dozen U.S. cities without any major incidents," using *only* their camera-based (vision-only) system.
7. **Waymo's Incidents:** In contrast, there are "multiple incidents now catastrophic major clusterfucks from Waymo," despite Waymo vehicles being equipped with an extensive array of sensors including Lidar, Radar, "gator sonar," and "every other form of communication and sensor on the planet" (including a sarcastic mention of "smoke signals").