Some people think so. I interviewed someone who, on a screenshare, would just type every question I said, verbatim, into antigravity. Then he'd look at the output for a second and say "Hm this looks good" (it was not) and then run the code and paste the error back into the prompt. It was a surreal experience. I didn't end the interview early because it was so incredibly wild I couldn't even believe it. I don't think he had a single thought the entire time that wasn't motivated by the LLM output.