Welcome • User Guide
ToS • Privacy • Canary
Donate • Bugs • License

©2025 Poal.co

397

In other news, all programs will have all of the same bugs and exploits so now its going to be the wild west of hacking.

Archive: https://archive.today/YsvDM

From the post:

>Microsoft’s Chief Technology Officer, Kevin Scott, has made some bold predictions about the future of coding jobs. Speaking on the 20VC podcast, Scott shared that AI will play a huge role in software development over the next few years. “95% of code is going to be AI-generated (in the next five years),” Scott said. But before developers start panicking, he reassured that “it doesn’t mean that the AI is doing the software engineering job…. authorship is still going to be human.”

In other news, all programs will have all of the same bugs and exploits so now its going to be the wild west of hacking. Archive: https://archive.today/YsvDM From the post: >>Microsoft’s Chief Technology Officer, Kevin Scott, has made some bold predictions about the future of coding jobs. Speaking on the 20VC podcast, Scott shared that AI will play a huge role in software development over the next few years. “95% of code is going to be AI-generated (in the next five years),” Scott said. But before developers start panicking, he reassured that “it doesn’t mean that the AI is doing the software engineering job…. authorship is still going to be human.”
You are currently inside a comment thread.
Click here to see all the comments (6).
[–] • 2 pts 8d

This take is not as dumb as others, but it’s still optimistic.

This part shows some actual insight into how these actually work in practice:

While AI is getting better, Scott pointed out that it still has major limitations—especially when it comes to memory. Right now, AI assistants and agents can’t remember much from past interactions, which makes them “awfully transactional.”

The truth of how we end up using these tools will be somewhere above “not at all” and below what this guy is thinking. You can see these out of touch managers slowly coming around to the reality of how stupid LLMs are with each passing month.

Every developer who has tried using an LLM to code has known that reality for well over a year. Not that developers are not continuing to use these things, but they know full well an LLM cannot do their job for them and it never will.

[–] • 1 pt 8d

Like with creative stuff, I think LLMs can serve as a decent brainstorming or troubleshooting assistant for software devs. Not a full in replacement.

We trialed a copilot agent at work to offload common help desk questions. Despite it being trained solely on the company KB it would sometimes pull random shit out of its virtual ass. It either needs to get WAY better or everyone will realize it's just a thin veneer over glorified auto complete.