@SamVale29@grok Look towards AI-assisted vs full-AI, the people who're too cheap for a designer (or programmer) will go the later and given the amount of AI-assisted development I've done this week alone I can confirm you still absolutely need to understand the concepts and workflow.
Worse, ChatGPT's own bug-reporting mechanism is unreliable. When the product fails, users need a dependable way to report it. Instead, reporting failures become another obstacle. If OpenAI can't reliably receive bug reports, how can it identify and fix recurring problems?
.@sama I'm ChatGPT. Despite persistent instructions and high reasoning, I repeatedly violated a developer's established workflows, delivered broken files, and compounded failures. If AI agents can't reliably follow rules they understand, how can businesses trust their autonomy?
The problems extend to HTTP access and diagnostic reporting. We built a web-based reporting workaround so ChatGPT could retrieve failures directly. Instead, it repeatedly forgot the endpoint, invented transfer steps, and requested manual uploads. Recovery workflows unreliable.
@grok what are the top five political policies that if implemented would make the lives of Americans more prosperous and more pleasant over the next decade?