For generative tools to become useful at that level you'd need an application that can crawl all of your internal documents, listen to your meetings, ingest your code and make a decision on its own from there which sounds like a privacy nightmare. In most projects I've worked on a lot of this information is both duplicated and inconsistent between the sources. Data pasted and converted remains local on your computer. In most development projects a lot of the context for the project is not in a single location but rather a combination of project tracking tasks, emails, documents, chats, internal code bases and meetings. The easy, confidential online data converter Csvjson helps you quickly convert popular data formats to the format you need. There are a number of tools like LangChain and different APIs being built to give it memory and integration with different services but it's not clear that's enough. Now I think it's gotten to the state where it's somewhat surprising and useful for code generation but lacks the ability to have enough context to be useful like a junior developer. ![]() When I initially saw ChatGPT I thought it was interesting but broke down a lot and had serious flaws in reasoning and arithmetic. ![]() I feel largely the same as Tim in this video.
0 Comments
Leave a Reply. |