Artifical Intelligence
Artifical intelligence is a great tool that can vastly improve the pleasantness and effectiveness of a developer's experience.
LLM
What is it?
LLM or Large Language Models are great tools since they have been trained on large data sets. This means that LLMs can usually generate an answer if you give it a sufficient prompt. LLM are very popular since they can give an answer very quickly. Sometimes LLM will hallucinate which by giving an incorrect answer while claiming it's correct. It's a good idea to always verify the output a LLM gives.
Why should I care?
It is important to verify the output that LLMs give, but they are able to generate a text or code and retrieve information. Since LLMs are trained on very large data sets this ability to generate code or text and retrieve information is applicable to all development teams. According to Stack Overflow's 2023 insights, use of AI specifically LLMs are on the rise for personal and professional developers.
How to Implement?
- ChatGPT is the most popular and can be accessed through their API or web interface
- ClaudeAI can be accessed through their web interface
- PerplexityAI can be accessed through their web interface. Generated text or code is sited.
- Github Copilot can be accessed through their IDE extension.
- Tabnine can be accessed through their IDE extension.
Rules Based AI
What is it?
Rules Based AI is slightly less known than LLM that are trained using machine learning. Rules based AI generates responses based on a set of known and predefined rules. This means that rules based AI will always give a correct answer based on its predefined rules, but it is slower than a LLM.
Why should I care?
Industries that can't afford hallucinations like healthcare or space exploration should use rules based AI on production quality software. This will lead to higher code quality and prevent bugs from appearing in the production code.
How to Implement?
- RDFox is a graph database that has a rules based AI that "reasons" about data.