Good reminder. Personally it probably makes sense to learn write software when using tests actually makes coding faster. It is possible most of the time, and 30% test coverage is better than 0, and 80% being better than 30%. The 100% can be hard to achieve (there are deminishing returns there) in a general case. Choosing the shape of the "test pyramid" is another hint. Even sacrificing unit tests for wider tests can be better than nothing.

I believe the real problem is that many developers just do not know how to write tests. If it's not everyday practice, a second nature, of course it is slow.