You're asking the right question: "what can't AI do?" and I also wrote an article on this topic: https://joshuaamathias.medium.com/humans-value-beyond-ai-b7e6a99b2093

But this article sounds like it could have been generated by AI because it's describing things that people have typically associated as hard for AI to do (your brain, like everyone's, was conditioned or "trained" by society to think certain ways) and it sounds convincing, but it's just an opinion with no data or concrete science to back it up. That's harsh but I'm just trying to make a point: to truly understand what humans can offer that AI can't we need to go a level deeper in detail and have a balanced discussion.

For example, on your 3 points:

1. Contextual Awareness: What contextual awareness can be done by an AI vs what can't? All 8 contextual factors you listed can be accounted for by an AI as they're just about information, rendering the whole point meaningless. You could strengthen the argument by discussing specific situations where an AI might not have access to certain contextual information, but then that's just a matter of solving the access problem, e.g. a camera to see, physical touch, etc.

2. Conflict Resolution: Why can't it "really" help us decide? Like any experiment, it's important to consider the baseline we're comparing against: Humans. Do humans make reliable, effective, unbiased decisions? Given all the information at its disposal on the previous decisions of millions or billions of humans, AI has a chance to do better. Now, where AI might fail is by being more likely to make mistakes that are /very/ wrong in certain circumstances where a human wouldn't. That's an example of going a level deeper, but this still seems like a solvable problem through analysis, guardrails, and automated checks.

3. Critical Thinking: Problem-solving, curiosity, creativity, inference and strategy. Could we design a system to copy the way humans do these things and apply that to new tasks and new idea generation? I don't see why not. Just because something is trained on the past doesn't mean it can't do something novel. Our brains are trained on the past too.

The best argument I've heard so far for what humans can do different than AI, other than spiritual/psychic abilities and awareness independent of the body, is that humans have a body that itself is a living organism and provides information to our brain in a way that's extremely difficult for an AI to simulate. Now, is that something needed to perform most jobs? Maybe not.