I don't get this article, it's clearly got a bone to pick with self-checkout and seems to be contradicting itself in the process:

Consumers want this technology to work, and welcomed it with open arms. [...] In a 2021 survey of 1,000 American shoppers, 60% of consumers said they prefer to use self-checkout over a staffed checkout aisle when given the choice

Okay, so even given the myriad of poor implementations out there, a majority of people prefer it. But then at the end:

Simply, "customers hate it".

Oh really? Because your quoted survey seems to say the opposite. And then there's stuff like this:

In addition to shrink concerns, experts say another failure of self-checkout technology is that, in many cases, it simply doesn't lead to the cost savings businesses hoped for. Just as Dollar General appears poised to add more employees to its check-out areas, presumably increasing staffing costs, other companies have done the same.

This is too light on data. Even a luxurious 1 cashier per 2 self-checkout stations will result in large cost savings for a business where employee costs are a significant fraction of total expenses. Especially in low margin businesses like grocery stores, removing even small amounts of overhead makes a big difference. Just because stores are adding a few employees back, doesn't mean cost-savings are completely negated.

Despite self-checkout kiosks becoming ubiquitous throughout the past decade or so, the US still has more than 3.3 million cashiers working around the nation, according to data from the US Bureau of Labor Statistics.

Surprise, a large nation did not completely get rid of cashiers! The number is meaningless without more context: did the number of cashiers go down? What about average cashiers per store? Where is the data?

My point is, maybe companies just went too hard on the cost-cutting and are trying to find the right balance. What is the best ratio of self-checkout to classic cashier checkout? What is the right amount of self-checkout assistants? How do we make checking out yourself a good user experience? All of these things are still being experimented with. What does seem to be clear is that self-checkout has become near ubiquitous, and therefore it is most certainly not a "spectacular failure" by any definition.