I have said this before, I will say this again: we've been using ML for pricing changes for different groups of users on ecommerce sites for literally decades.

Most simplistic example:

1. User arrives on site; we believe we haven't seen them before (their cookie is new and based on our ID graph, they may very well be) and they're behavior on site (e.g., how they navigate it; what they click on; how they scroll) is indicative of an entirely new user. For this user, we give a discount to try and capture the sale and keep them coming back with nurture activities (e-mail or retargeting).

2. User arrive on site; we KNOW they are a returning user (e.g., we have their cookie tied up to pre-existing rows in the ID graph; they came in from a retargeting ad; and/or their on-site behavior is indicative of a returning user). For this user, they get the 'normal pricing'.

---

The only difference here is that the article claims they use AI and AI is scary.