This article, Datasets Over Algorithms — Space Machine, makes a good point, namely
…perhaps many major AI breakthroughs have actually been constrained by the availability of high-quality training datasets, and not by algorithmic advances.
Looking at this chart they provide illustrates the point:
I’d argue that it isn’t solely datasets that drive A.I. breakthroughs. Better CPUs, improved storage technology, and of course new ideas can also propel A.I. forward. But if you ask me now, I think A.I. in the future will need better data to make big advances.