Positive users agree AI code generation now amplifies expert skills and productivity returns, while negative users reject the shift as a myth or revisionist history that overlooks benefits to juniors.
Based on 15 visible X reactions from 93 accounts; directional sample.
Ask a question below.
Published answers will appear here.
Absolutely agree that the flip already happened, but I’d add that if code gen is not the product of your work (software dev), but rather a tool to get to the product, you probably have been drowning in magic of these power tools since at least ~Opus 4.0. For instance, for data science projects, you don’t really care about efficiency and stylistic details on many stages, and so ai generated charts, data collection, etc. work in 99% of cases and you can focus on experiment design and other fun stuff.
@fchollet Claude 4.5 models are still used and available in enterprise. The quality of their output can't be said to be that much different. You guys seem to become unwilling or unconsciously (hopefully) bullshitters.
@fchollet I don’t know about that, I’m building things now that I was never capable six months back.
@fchollet Completely agree.
Absolutely agree that the flip already happened, but I’d add that if code gen is not the product of your work (software dev), but rather a tool to get to the product, you probably have been drowning in magic of these power tools since at least ~Opus 4.0. For instance, for data science projects, you don’t really care about efficiency and stylistic details on many stages, and so ai generated charts, data collection, etc. work in 99% of cases and you can focus on experiment design and other fun stuff.
@fchollet Claude 4.5 models are still used and available in enterprise. The quality of their output can't be said to be that much different. You guys seem to become unwilling or unconsciously (hopefully) bullshitters.
@fchollet I don’t know about that, I’m building things now that I was never capable six months back.
@fchollet Pivot has occurred!
@fchollet disagree.
The weak AI code gen we had until late last year was most useful to low-skill programmers -- it was raising the floor. It was essentially useless to high-skill programmers -- you could move faster and ship better code without. This has been completely flipped: the strong AI code gen we have now is *most* useful to high-skill programmers, while low-skill programmers are vastly underutilizing it or sometimes drowning in it. It went from a crutch to a power tool.
Positive users agree AI code generation now amplifies expert skills and productivity returns, while negative users reject the shift as a myth or revisionist history that overlooks benefits to juniors.
Based on 15 visible X reactions from 93 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@fchollet disagree.
The weak AI code gen we had until late last year was most useful to low-skill programmers -- it was raising the floor. It was essentially useless to high-skill programmers -- you could move faster and ship better code without. This has been completely flipped: the strong AI code gen we have now is *most* useful to high-skill programmers, while low-skill programmers are vastly underutilizing it or sometimes drowning in it. It went from a crutch to a power tool.