Because machines operate in average, what they serve might work for people close enough to that average.
So when going to an automated doctor for a headache, or a designer bot making you a logo, if you’re close enough to the median taste (or unable to articulate your preference) such services shall seem good, and affordable.

However once you start asking follow up questions to that diagnosis, or inquire about a typeface choice – a machine (by today’s definition and under its current architecture) will be unable to help.

But most of us seek to understand the service or product we are paying for. We want to make sure we’re making the right choice with our money, and want to be able to address future situations without having to call the ‘shop’ again every time.

Numbers will need to be studied, but I argue that this median offering will only catch a small portion of the market, and that the majority of us will still need a human to contextualize, explain and empathize the process.

Promises of full automation have a tendency to shift to a collaboration where a thoughtful human translating between a fast computer and a costumer (e.g. live chat to a person on an Ecommerce website, Adwords sales people).

We are going to see a lot more examples of post standardization, situations where consumers are actively pushing back on median preferences, and are looking for small batches, custom or otherwise instantiated services.