Excellent point, Dave! And the white hat hackers in the domain in thinking is what we need. As passionate and dedicated AI researchers may be, they are akin to the designers and builders of cars. Very few of them become F1 race drivers, and they need data from all users of their product to fine tune it. Some of the best data will come from the average user for improvements of a car model. But the extremes of what different components can handle (tires, suspension, material strength, etc.) comes from uncommon and extreme testing of these...

The AI models like ChatGPT decided to give a car to everyone and the feedback from users will help define the guardrails, the road rules, and how and where we need to apply brakes. When the car was first built, there were no paved roads, no crosswalks, no light signals, no signs... etc. Would they ever become necessary if we simply stopped building, driving, and testing cars?

We need white hat thinkers to use ChatGPT and run thought experiments and share them... This is how the 'data' of what they can do can truly be put to public scrutiny and not just sit on the servers of AI companies. ChatGPT public release for free could be one of the most important event in the history of human-AI relations...

As someone testing GPT models before ChatGPT hit the scene, I wrote about it in a few blogs about thinking, here is one of them: https://gesikowski.medium.com/deep-thinkers-wanted-be-original-6fb4c8721a15