Many of the mathematical models on which Artificial Intelligence, AI is based were formally called statistics/econometrics, now they are also called machine learning.
The problem with these mathematical, scientific models is that they are tailor made and hence difficult to standardise and scale and subsequently sell as a commercial software product. This is the reason why firms have not been hiring on constant payroll statisticians, modellers and econometrician. That is, until 9 years ago, when in 2013 Artificial Intelligence exploded.
Companies like Palantir Technologies Inc or simply Palantir are an example how Artificial Intelligence and big data analysis can be scale. 1 year ago Palantir was valued at more than 40 billion USD. Actually, much of the work done by Palantir Technologies is still project based, that is tailor made, but also a large part of Palantir's revenue comes from scaled artificial intelligence software products.
Actually, the problem is on a deeper, scientific level. The problem with all those statistical models is that they are linear and rely on staff like p-value, t-statistics and so on, which also have a linearity problem.
Translation, one cannot be 100 % certain, that the statistical relationship the model shows is not due to a fluke, luck.
So Artificial Intelligence, AI can be disrupted, I think, if a "genius" makes a breakthrough first in the scientific methods underlying statistics/econometrics/machine learning/Artificial Intelligence, AI models and also a software coding quantum leap.
Easier said than done, but the company that does that could become the world's most valuable company, because Artificial Intelligence or AI is the future.