Maybe your work is just in the early stages of exponential growth?

AlphaGo -> exponential improvement

GPT -> exponential improvement

FSD -> exponential improvement

Cost to train -> exponential decrease

Compute performance of AI chips -> exponential increase

If you think that an AI that took 10 years to be 10% as good as a coder will take another 10 years to be as good, who is delusional?