1. In the same amount of time, AI now generates about 5× more code than before.
2. In the same amount of time, engineers have to review about 5× more code than before.
3. The time spent reviewing code is now greater than the time spent writing it.
Conclusion: In many companies, AI has actually reduced the overall efficiency of engineering teams.
If companies really want AI to improve productivity in production environments, the key is to improve the quality of generated code and the efficiency of the human-in-the-loop validation process.
What teams actually need to find is a balance point, and that balance point depends heavily on the methods used by individuals and teams.
At that point, AI might generate only 2× more code instead of 5×, but the time required for both individual and team code reviews could drop significantly. In that case, code quality would not decrease—in fact, it could improve—while iteration speed would still increase in a meaningful way.
2. In the same amount of time, engineers have to review about 5× more code than before.
3. The time spent reviewing code is now greater than the time spent writing it.
Conclusion: In many companies, AI has actually reduced the overall efficiency of engineering teams.
If companies really want AI to improve productivity in production environments, the key is to improve the quality of generated code and the efficiency of the human-in-the-loop validation process.
What teams actually need to find is a balance point, and that balance point depends heavily on the methods used by individuals and teams.
At that point, AI might generate only 2× more code instead of 5×, but the time required for both individual and team code reviews could drop significantly. In that case, code quality would not decrease—in fact, it could improve—while iteration speed would still increase in a meaningful way.