@davis7@shivamhwp Isn’t that just a matter of training data, tools, context and model capabilities? Isn’t it better to build for the models in 6 months from now, than for the models of today?
@GergelyOrosz It is a unique thing to LLMs, the closer the computer gets to being human, the more fallible it is. Humans make mistakes, and so do LLMs. Now, more than ever, do we have to question a program’s output.
@GergelyOrosz I used this app extensively during my university days, and the team was even nice enough to give me a student discount and I can confirm it’s the best tool to do anything pair programming related, towards the end I was even using for regular calls
@joyofcodedev I love using that library. I used it to host an internal knowledge share at work and got really great feedback about the visuals and interest in what i used to create them.
@davis7 I created a custom GPT for that and provided the full docs for it. Worked really well. Even gave it the markdown of all documentation and the implementations of all shadcn-svelte components which worked reasonably well but it still used native html elements sometimes instead