I've been using Github Copilot, and I've been enjoying it. When I'm writing most code, it does okay, but doesn't really help too much yet. But every once in awhile it shines.
For things that are of less importance to me, it really does well. Testing is one of those topics. It is good at adding a lot of boilerplate copy.
Seeing this bafflegap reminds me of the war memorandum by Winston Churchill in 1940 concerning brevity. There is a lot to be said for it. But unlike ink or computer memory, brevity is in short supply.
In its current form, I see Github Copilot as useful for boilerplate-driven tasks such as testing an infrasturcture-as-code. As AIs expand, I expect to see more content writen by AIs. I see this already in Notion and other tools.
There is a downside to these tools. We now have the ability to write copy that is writen by no one, and read by few for the sake of ranking on automated search engines. We also have the ability to have machines write directions that are intended to be only consumed by machines. Could it be that we are setting ourselves to drown in boilerplate?