I think what you're really seeing is the difference between 'photography' and 'computational photography'.
Nothing I can see in the first photo hits 0,0,0 or 255,255,255, it all fits quite neatly between the extremes without ever hitting them. I'd call that a pretty good exposure guess by the camera. The camera doesn't know what you're taking a photo of, or which parts you might like to emphasise. It captures light, the rest is up to you.
The second photo, on the other hand, has been post-processed by the phone to be far more pleasing to the eye. It does actually hit 0,0,0 in places & there's the occasional 255 in the blues, in the sky. It could only do that in post; nothing could be that accurate on the fly.
The phone, algorithmically kind of does know what you're taking a photograph of. It's been programmed to recognise many aspects of many scenes & what to do to make them look 'nice'.
In short, the 'photograph' has left enough room to do your own post-processing. The 'computational photograph' has done it all for you… whether you like the results or not. The developers have worked on these algorithms a long time, to try to ensure that most of the time people do actually like it. tbh, I still think it's a bit 'thin' & unenthusiastic.
By extreme contrast, here's a quick go at brutally shoving the first image in the direction of "Hey, isn't this what we expect a sunny day to look like?!" in a sort of over-enthusiastic picture postcard kind of way, without actually trying at all to copy what the phone did. Consider it merely a third way of looking at it.
