I’m a C# .NET developer with extensive front-end and Azure cloud experience. I am also an expert on LLMs, and know how to solve virtually any problem associated with language models. I have thousands of hours of experience under my belt since GPT-2.
Great to hear! I think as more powerful local models become available there will be a strong demand for tooling that can match functionality like 'GPTs' and Claude Artifacts on-prem.
The reason AI will quickly replace programmers is because of one major reason that many do not consider: AI allows you to build entirely new types of applications which will not require the same software patterns we use today. In fact, using AI, you can create software without 'code' (in the traditional sense). Until you start building genai apps and see the new paradigms at work, it will be difficult to perceive.
If you haven't checked out my project, Cheevly, you should look into it. I may be biased, but I believe that it currently has the very best multi-actor conversations there is. It's free, but requires a bring-your-own GPT key.
Cheevly is my AI desktop assistant which is releasing within the next 30 days. It features custom voice, real-time vision, desktop automation, multi-character roleplaying, and many other features. In this clip I'm playing World of Warcraft while it roasts me in Trump's voice. It's a work-in-progress, with some glitches in the voice. Once OpenAI does not have such strict rate-limits, I'll be able to make significant enhancements to the capabilities.
It can take weeks for project managers to realize that a programmer misunderstood the requirements in a deliverable. It will take seconds for managers to realize that AI misunderstood the requirements. Managers will be able to rapidly iterate on requirements in ways they've never been able to in the past. And it just so happens that distilling requirements is actually something that AI is really good at.
How would AI in it's current state support gathering and communicate requirements?
In my experience, it requires careful observation, an investigative approach and deep reflection to understand one's needs, context, goals, etc, in order to translate into software requirements.
I'm a big fan and user of GPT for coding and everything, but I don't see it helping a lot with requirements in the foreseeable future. Not with its current capabilities...
I'd guess the end user (product owner) of the software would interact with some sort of ChatGPT and describe what they want/need and we'd iterate from there. At the very least it would be a good support tool for initial scoping.
ChatGPT would also be useful for the developers to learn more about the domain (assuming no hallucinations) before the kickoff meeting. Let's say you're building some sort of accounting software for the banking sector. You could have a chat session that explains how business processes work in that area, what regulations apply etc. You'd also pick up the correct words/phrases that are used in that domain etc.
Remote: Yes
Willing to relocate: Maybe
Technologies: C#, Javascript, Cloud, LLMs
Résumé/CV: https://www.linkedin.com/in/cheevly
Email: josh.d.griffith at gmail
I’m a C# .NET developer with extensive front-end and Azure cloud experience. I am also an expert on LLMs, and know how to solve virtually any problem associated with language models. I have thousands of hours of experience under my belt since GPT-2.
reply