Way to trigger FOMO. Not wrong either. That said, consistency in quality of models is a real issue. Even the big LLMs seem to have a different head on over the weekend vs the week. If you were to test them for quality every time you use them, you’d run out of credits before you’re done. Many days it feels like the models cause more mess than they solve (and there is no leader since they all do this so you play tilt between one company and the other). That’s why I have hope for local models and keep searching among the new releases. Trusting any model however beyond a clean slate new build is a fool’s errand.
The people with skill are still very very needed. I think the bigger bonanza will be for skilled engineers to clean up the messes made by LLMs and the massive critical failures that will happen because management trusts LLM code when they don’t really know it, for the savings.