Originally posted by UncleVanya they use computational photography tricks to enhance the way the images look. This can be fantastic when your vision of the image matches that of the creator of the enhancement - not always the case.
This is really the crux of the matter for me. If we could see the uncooked, unprocessed raw image from even the latest phone camera sensor at 100% reproduction, it would be
garbage compared to the raw from a modern DSLR, especially for shots taken above base ISO.
Instead, the phone cleverly delivers an image that has been automatically and pseudo-intelligently subjected to mind-boggling amounts of processing and computational transformation - in real-time and post-capture - to produce something it thinks will look impressive at modest reproduction sizes... all with little effort or skill on the user's part; which is perfectly fine, as that's really the whole point of such devices.
For a reasonably broad but nonetheless quite limited set of use cases, the latest phone cameras will produce great looking images at modest reproduction sizes, with little or no effort. Indeed, to produce the same results at the same reproduction sizes with a DSLR will usually require considerably more knowledge, skill and effort from the photographer.
Where DSLRs excel, though, is in the range of use cases they can handle, and the potential for excellent image quality at larger reproduction sizes. Of course, leveraging these capabilities requires - you guessed it - knowledge, skill and effort... but the reward is the satisfaction of having applied one's craft to produce a technically and artistically competent result, rather than having a machine give you what it thinks you wanted.