80% is a pretty abysmal success rate and means it's very unreliable.

It sounds nice but it means at least 1 in 5 are bad. That's worse odds than rolling 1 on a d6. You'll be tripping over mistakes constantly.

  > 80% is a pretty abysmal success rate and means it's very unreliable.

EXACTLY. imagine if your car would at best start 80% of the time...

At least if the car doesn't start, you can immediately tell.

Imagine Google Maps driving directions took you in the correct direction 80% of the time.

From the study: " [80% is] the same level of agreement between humans."

So if 80% is abysmal, you are getting the same abysmal level from human code reviewers.

Not the same thing. With human code reviewers, you can talk to them and resolve conflicts. You can even ask them to review code differently in future.

The whole point of a tool is to reduce workload. If you have to frequently correct it, then it's not worthwhile to use.

LLMs will not have the context behind the lines of code in the CR.

Sure there's no bug with how the logic is defined in the CR or even in the context of the project, ti maybe won't throw an exception.

But the LLM won't know that the query is iterating over an unindexed field in the DB with the table in prod having 10s of millions of rows. The LLM won't know that even though the code says the button should be red and the comments say the button should be red, the corporate style guide says red should be a very specific hex code that it isn't.