Well its nice to see that the constant moaning about AMD's GPUs will just be replaced with AI discussions.
I'm personally gung ho and all on board with AMD abandoning high end GPUs. I think they should abandon dGPU add-in cards entirely as its a pointless waste for them. They should have pivoted to eGPUs, and have been moving towards large dedicated gaming APU boxes. This would let them focus laptops and tablets on efficiency, and for gamers it'd have benefits as well (CPUs close by lowers latency, shared memory and even cache so something like 3D V-Cache could benefit GPU more directly than it currently does in the X3D CPU). It would have utilized their embedded/semi-custom teams, but also they could use the entire market chain used for making dGPUs for those so it wouldn't even need a huge shakeup in how they do things since dGPUs are already large enough, would be almost trivial to just make it a fully enclosed box. They also could use it as an opportunity to change the form factor to benefit things (a Gamecube sized box which is basically just a shroud for a modern tower cooler would improve cooling and noise compared to the 3 fan shrouds of modern dGPU). They could do some other further things, like putting an x16 PCIe port (no longer needed for GPU) for ultrafast SSDs where the bandwidth would be more like DRAM memory bandwidth but with improved latency. Uncompress entire game assets and stream them in at high quality, reducing load times and also pop-in. Maybe they could even pre-process ray data for path tracing or something that would make that more feasible.
It would also let Microsoft and Sony (rumors are that despite its success - Playstation was basically singlehandedly keeping Sony overall in business in the 2010s as they were floundering in a lot of their other markets like movies and TV and other consumer electronics and PCs and smartphones - that they struggle with accepting the development cost of Playstation hardware) get out of the costly hardware development business, but keep the benefits of all the software development (run the Xbox software environment on these boxes, and of course Steam could be a viable option as well - Sony really could do themselves favors by partnering with Valve there and it would benefit gamers and developers as well).
That's step 1. Step 2 they replace your beach with a nicer beach, and Step 3 they replace your friends with hotter friends.
You're assuming those are his friends...
And the moon in that pic is perfectly fine for what that pic is. The problem is the person is taking issue with the wrong thing (the poor light/light source handling of smartphone cameras that is a result of the necessary - due to it being a smartphone camera - lens and sensor size situation). Frankly their skills at using even a smartphone camera might be the biggest issue with the quality of that pic. They certainly either are trying to pass off a still capture from a vertical video or did a crop (of a highly zoomed pic) or don't know how to use the resolution of their camera sensor properly. Heck I think you can buy lenses (or filters) that could be used on a smartphone that would have improved that pic for them. Using AI to replace the moon (which, uh, the inherent problem is calling this stuff "AI" when its not at all the AI that people think of as AI - which none of this "AI" can even do and they're just straight up lying about that fact; machine learning image modification is neither new nor really needs these overwrought generative AI models to do) definitely doesn't make it any truer or whatever nonsene argument is being used to justify that.
On the flipside, smartphone cameras inherently are manipulating every image they take (unless you're specifically not doing that which I don't know of basically anyone, including pro photographers, that even use them like that outside of when making something specifically for that reason, and most pro photographers talk about how they use their smartphones specifically to not have to do a lot of work to get an acceptable quality picture), so complaining about using it to help overcome major deficiencies that are inherent because of technical limitations is well, already too late. I would like if it keeps the native image info and you could do a quick switch to see the original vs the modified version. I wouldn't even mind a pro mode where it could do better tweaks of like the color space, making the pro features more useful. Maybe even make an app that teaches people what different settings do to impact image quality (which could help them take better pictures), and then make an automated process (think almost like an eye exam "do you prefer image 1 or 2 and it cycles through a bunch building a profile) so people can create their own algorithm to adjust images to what suits them (and could then be used across apps vs say Snapchat filters and the like).
And just watch, this "AI" replacement or removal will cause all sorts of drama, just like the iMessage stuff does. Actually it already is because people are insane and have gone full creepy, making nudes of people and then sharing that for gross or mean reasons. Which, wonder how many people will use it to add in celebrities they like to make it seem like they were taking pics with them. Bet scammers will use that. "I'm totally rich see me hanging out with these other rich and famous people!"