Every time I read something along the lines I have to wonder whose code these people review during code reviews. It’s not like the alternative is bulletproof code.
There's no such thing as bulletproof, but there is definitely such a thing as knowing where your vital organs are and how to tell when they've been hit.
And the others are going to be replaced by one engineer and an AI of equivalent caliber.
1. Many developers are currently employees who objectively produce code of equal or lower quality than AI as it currently is.
2. It is cheaper and more productive to higher fewer, more competent, people and replace the less productive ones by AI (possible due to 1).
3. Short of regulations preventing it, companies will follow through on 2.
what can be asserted without evidence can also be dismissed without evidence
Considering 80% of team mates are usually dead weight or mid at best (every team is carried by that 1 or 2 guys who do 2-3x), they will do the bare minimum review. Let’s be real.. PIP is real. Job hopping because bad is real.
It’s a problem. I have dealt with this and had to fire.
Anything coded by an LLM risks being under-generalised.
Asking an LLM to think in a generalised way does not make it an AGI. The critical ability to generalised beyond learned patterns and to not only come up with arbitrary patterns but to use correct logic to derive them is missing from LLMs because LLMs do not have a logical layer, only a probabilistic one with learned constraints. The defect is the lack of internal logical constraints. It’s a big subject.
I say more about it here:
https://www.forbes.com/sites/hessiejones/2025/09/30/llms-are...
Aka
“layered system”
Should, yeah. But that was not true even before LLMs.