Seán Fenian“When #AI Became a #Force #Multiplier for #Incompetence”
https://techtrenches.substack.com/p/the-great-software-quality-collapse
When:
• Apple Calculator leaks — not uses, leaks — 32GB of RAM
• Windows 11 updates regularly break the Start menu
• 16GB of memory consumption for 50 tabs in Chrome is “normal”
• macOS Spotlight wrote 26TB to SSDs overnight
• VSCode leaks 96GB of RAM through SSH connections
• A single CrowdStrike configuration file missing one array bounds check crashed 8.5 million Windows computers globally
And more.
Still we're relying more and more on “AI” (STOP CALLING IT THAT, LLMs are NOT intelligent!) coding. Because you can hire someone to tell an LLM what you want it to do and hope for the best, generating code that its own “developers” don't understand and, critically, cannot test or debug. They just have to take the LLM's word for it that the code is correct, when we already know that the LLM will lie to protect itself. LLM-generated code contains 322% more security flaws than human-written code, and 45% of LLM-generated code contains exploitable flaws.
After 12 years in engineering management, the pattern is unmistakable:
Stage 1: Denial (2018-2020) “Memory is cheap, optimization is expensive”
Stage 2: Normalization (2020-2022) “All modern software uses these resources”
Stage 3: Acceleration (2022-2024) “AI will solve our productivity problems”
Stage 4: Capitulation (2024-2025) “We'll just build more data centers.”
Stage 5: Collapse (Coming soon) Physical constraints don't care about venture capitalAnd the entire model of the tech industry as it now stands is 'optimizing' out the pipeline that creates developers who know what they're doing. So this is only going to get worse, unless we turn around before it is the blind leading the blind.
Assuming it's not already too late.
Jilicc Bookmarker
TG777 Game
BetSo88
Radio Azureus
acemaxx
Rich Holmes
Ronnie / Rekkerd.org
Rich Holmes