Whether or not markets will be replaced by centralized planning is a separate story. But it seems like improvements in artificial intelligence will help solve the primary problem with centralization, which is that markets are more "intelligent" and efficient insofar as they enlist the distributed, skin-in-the-game intelligence of individual participants. State-run economies, on the other hand, not only operate with an extreme information deficit—they also run smack-dab into deep, intractable bureaucratic/managerial issues (perverse and misaligned incentives, surrogation and reportage lossiness, everyday corruption).
Having a computer (or more realistically, many computers) run an economy comes with its own slough of difficult design problems, not least of which is a hard alignment problem. Moreover, these computers would have to handle their own data collection and tagging—otherwise all the aforementioned reporting/management issues rear their head again. This is not possible given current AI paradigms—there would have to be a serious shift in how we build and train the technology. Not impossible, but not happening til the next century I'd bet. At this point, centralized economies could become tenable.