This is the thing I’m trying to advocate people to understand. It’s not that they were wrong about AI a year ago - AI has changed in radical ways people couldn’t confidently predict within just a year.
The AI coding agents a year ago DID create slop that would dillute your code base. The AI coding agents today generally don’t.
It’s becoming increasingly less valuable to combat your LLM and scrutinize it’s output. You can provide more market value by assuming it’s 95% correct, doing some manual verification, and architecting it in a way that leads to low integration complexity. The OOP principles are still relevant - abstract, encapsulate, etc.
I swear I’m not an AI bro despite the above comment being somewhat pro-AI. I’m just trying to be pragmatic about the state of our field.
The standard form is “I hate AI as much as the next person, but–” followed by whatever you were going to say anyway. It seems to appease the identarians.
Maintainability became the go-to jibe this year. What a high-level requirement to complain about. Like sure, generated code can demonstrably work, and read sensibly with comments and tests, and interoperate with other programs… but in ten years you’ll be sorry! As if local models two years from now won’t make these behemoths look primitive. The bubble could pop tomorrow and it’d only increase pressure toward smaller and faster experiments by people with one-comma budgets.
A recent Computerphile video amounted to nervous laughter about chain-of-thought vanishing as a worthwhile glimpse of the model’s state. (On the shaky assumption it ever was.) At the end they mention prompts where the model gets to output one word as an immediate answer. Last year the kinds of problems they could solve “between words” was comparable to what a person could think through in thirty seconds. On current frontier models it’s more like half an hour.
This is the thing I’m trying to advocate people to understand. It’s not that they were wrong about AI a year ago - AI has changed in radical ways people couldn’t confidently predict within just a year.
The AI coding agents a year ago DID create slop that would dillute your code base. The AI coding agents today generally don’t.
It’s becoming increasingly less valuable to combat your LLM and scrutinize it’s output. You can provide more market value by assuming it’s 95% correct, doing some manual verification, and architecting it in a way that leads to low integration complexity. The OOP principles are still relevant - abstract, encapsulate, etc.
I swear I’m not an AI bro despite the above comment being somewhat pro-AI. I’m just trying to be pragmatic about the state of our field.
The standard form is “I hate AI as much as the next person, but–” followed by whatever you were going to say anyway. It seems to appease the identarians.
Maintainability became the go-to jibe this year. What a high-level requirement to complain about. Like sure, generated code can demonstrably work, and read sensibly with comments and tests, and interoperate with other programs… but in ten years you’ll be sorry! As if local models two years from now won’t make these behemoths look primitive. The bubble could pop tomorrow and it’d only increase pressure toward smaller and faster experiments by people with one-comma budgets.
A recent Computerphile video amounted to nervous laughter about chain-of-thought vanishing as a worthwhile glimpse of the model’s state. (On the shaky assumption it ever was.) At the end they mention prompts where the model gets to output one word as an immediate answer. Last year the kinds of problems they could solve “between words” was comparable to what a person could think through in thirty seconds. On current frontier models it’s more like half an hour.
I couldn’t agree more. There is a long list of valid reasons to hate on AI and the industry but AI = slop barely makes it onto the list anymore.