Especially in cases involving tech, it is important to have full access to evidence. So an independent analysis can be performed and additional investigation can be requested.
Very few people can understand machine code. Some can read small snippets, no one I know has written an application in machine code. There are compilers so they don't have to.
The problem is that a shared code base. Languages for verifiable code are not mainstream yet and probably too difficult for humans to learn. What software developers should do is define specifications and let AI create a verifiable imlementation.
Or improve the login procedure. It is really bad for company moralle if the organization is not able to recognize and act when it notices such a waste of resources.
It is a bit like self driving vehicles, there is really no self driving cars unless you take out the steering wheel. The only way to reduce white collar jobs is to reduce the number of computers, if you do not do that, then there will be a white collar worker feeding input and interpreting output of the AIs.
AI is now a commodity, and a cheap one, where the real value is how you use and apply it. That field haven't really begun yet, all we have seen is a fun chat show case and some fancy autocompletion. What we need is finding out how it can improve business and peoples lives. I very much hope for an AI robot butler at home.
True that Apple hasn't made anything interesting for a while. It is a consequence of aging in upper leadership with no ambition and a safe market position.
I hope that the entry of AI can remove the reliance in big screens, abondency of apps and excessive hours of device use. Perhaps subscription driven instead of ad driven.
Rectangular devices with round corners are the beige boxes of the 2020ies.