I like to use computers as a “proof of concept” type of device. In other words, full featured computers are a good way to implement a given application or feature, since it is already there, and capable of having the feature added by some relatively simple combination of software and specialized hardware. It’s a good way to test whether one would enjoy that application or feature, and live with it for a while. If it becomes part of the habit, then one can invest in the dedicated device.
I have also found that general purpose computers are easier to upgrade than single purpose devices. This is a broad generalization with many exceptions. This is one reason that I use the Swiss army knife of a general purpose computer first to implement a function or feature – often times the early versions of single purpose devices become obsolete - under powered or not fully compatible with the whole whatever it is – by the time that function or feature is mature and widely adopted. As a serial early adopter, I’d rather adopt early with a multi-purpose computer, then wait out the first generation of single purpose devices, and then when those devices become perfected, I’ll jump in.
I also do not see people giving up their smartphones any time soon and going back to single purpose devices like carrying a camera, watch, cell phone, reading tablet, etc. Smartphones are small but they are little general purpose computers, not little single-purpose devices.
Bottom line, there are different uses for each, and in some ways each represent a different stage of development of a given application. Two sides of the same coin. Both are necessary and will be for the foreseeable future.