I believe we do not truly have agency but have evolved to think and act as though we do. Since inputs to each choice are likely infinite (probably uncountable as opposed to countable), the lack of agency is difficult to observe.

I am largely concerned that the development and evolution of generative AI is driven by hype/consumer interests instead of academia. Companies will prioritize opportunities to profit from consumers enjoying the novelty and use the tech to increase vendor lock-in.

I would much rather see the field advanced by scientific and academic interests. Let’s focus on solving problems that help everyone instead of temporarily boosting profit margins.

I believe this is similar to how CPU R&D changed course dramatically in the 90s due to the sudden popularity in PCs. We could have enjoyed 64 bit processors and SMT a decade earlier.

Driving a manual is pretty easy, but it depends on how you handle failure. The learning process requires making a few mistakes while you become familiar with a clutch and shifting gears.

Also, I think it’s much easier to learn if you know what the parts are doing. Watch some videos on how it all works and what different RPMs sound like.