AI made writing code easier.
-
AI made writing code easier. It also made being an engineer harder. https://www.ivanturkovic.com/2026/02/25/ai-made-writing-code-easier-engineering-harder/
Early CodeGen enthusiasts talked about "I can accomplish code while walking my dog / taking my kid to the park". Shock and surprise, capitalism doesn't work that way.
Engineers are working longer hours, doing less fulfilling work, are expected to ship faster, often after their org experiences major layoffs, and the quality of output is lower.
-
AI made writing code easier. It also made being an engineer harder. https://www.ivanturkovic.com/2026/02/25/ai-made-writing-code-easier-engineering-harder/
Early CodeGen enthusiasts talked about "I can accomplish code while walking my dog / taking my kid to the park". Shock and surprise, capitalism doesn't work that way.
Engineers are working longer hours, doing less fulfilling work, are expected to ship faster, often after their org experiences major layoffs, and the quality of output is lower.
Engineers are also understanding the contents of their codebases less, leading to "cognitive debt".
https://www.media.mit.edu/publications/your-brain-on-chatgpt/
https://simonwillison.net/2026/Feb/15/cognitive-debt/"Technical debt" is where technical issues and mistakes pile up without being fixed. This is happening too, but "cognitive debt" is where more of your codebase has been produced without you knowing or understanding how it works, leading engineers often to become helpless in terms of being stewards of their own codebases.
And it's not just happening to software engineering. Workers everywhere are suffering from cognitive debt as AI rolls out across their workforces.
-
Engineers are also understanding the contents of their codebases less, leading to "cognitive debt".
https://www.media.mit.edu/publications/your-brain-on-chatgpt/
https://simonwillison.net/2026/Feb/15/cognitive-debt/"Technical debt" is where technical issues and mistakes pile up without being fixed. This is happening too, but "cognitive debt" is where more of your codebase has been produced without you knowing or understanding how it works, leading engineers often to become helpless in terms of being stewards of their own codebases.
And it's not just happening to software engineering. Workers everywhere are suffering from cognitive debt as AI rolls out across their workforces.
That paper is bullshit.
It's a preprint and they hide the "more"/"less" data in it.It's actually embarassing to use it to support your case, was that the strongest argument you found?
-
R relay@relay.infosec.exchange shared this topic
-
Engineers are also understanding the contents of their codebases less, leading to "cognitive debt".
https://www.media.mit.edu/publications/your-brain-on-chatgpt/
https://simonwillison.net/2026/Feb/15/cognitive-debt/"Technical debt" is where technical issues and mistakes pile up without being fixed. This is happening too, but "cognitive debt" is where more of your codebase has been produced without you knowing or understanding how it works, leading engineers often to become helpless in terms of being stewards of their own codebases.
And it's not just happening to software engineering. Workers everywhere are suffering from cognitive debt as AI rolls out across their workforces.
@cwebber I am frankly not looking forward to having to debug some LLM generated code and opening it up in my editor and seeing 2k lines of WTAF splatted in there.
-
AI made writing code easier. It also made being an engineer harder. https://www.ivanturkovic.com/2026/02/25/ai-made-writing-code-easier-engineering-harder/
Early CodeGen enthusiasts talked about "I can accomplish code while walking my dog / taking my kid to the park". Shock and surprise, capitalism doesn't work that way.
Engineers are working longer hours, doing less fulfilling work, are expected to ship faster, often after their org experiences major layoffs, and the quality of output is lower.
@cwebber Trying my best to avoid LLMs at my dayjob. Questioning if my tact of complete avoidance is sane under the pressure to adopt and "be more productive".
-
Engineers are also understanding the contents of their codebases less, leading to "cognitive debt".
https://www.media.mit.edu/publications/your-brain-on-chatgpt/
https://simonwillison.net/2026/Feb/15/cognitive-debt/"Technical debt" is where technical issues and mistakes pile up without being fixed. This is happening too, but "cognitive debt" is where more of your codebase has been produced without you knowing or understanding how it works, leading engineers often to become helpless in terms of being stewards of their own codebases.
And it's not just happening to software engineering. Workers everywhere are suffering from cognitive debt as AI rolls out across their workforces.
@cwebber@social.coop I'd offer that software development differs importantly from essay writing, and so it might not make sense to differentiate technical debt from so-called cognitive debt in that case."Technical debt" is where technical issues and mistakes pile up without being fixed.
Yes, but not just any issues, and not just any techniques. Technical debt involves issues that matter to the functioning of the codebase and the processes creating and sustaining it. Technical issues that have no current or future impact on the functioning of the codebase, the further development of the codebase, or the maintenance of the codebase, might as well not be there.
According to Peter Naur, this set of processes creating and sustaining codebases includes the "theory" of the codebase, which partly (largely) resides in the heads of the people who work with it. Naur argues that programming is theory building: if you (and your team) are not building a theory of a codebase, you are not programming, you are typing.
To sum that up, you cannot separate the technical aspects of code from the cognitive aspects; they are inextricably tied together. If there is technical debt, there is almost surely associated cognitive debt, and vice versa. Splitting the two apart is partly why we ended up in this mess: it allows space for the dangerous fiction that a machine can spit out code faster than people and therefore could/should be used to replace people.
