I mostly use the GPT-4 model. I have heard complaints about ChatGPT hallucinating form 3.5 users. I still think with the right context and prompt 3.5 can be useful. But good to be aware.

GPT 3.5 vs. GPT 4: What’s the Difference?

Preventing ‘Hallucination’ in GPT-3 and Other Complex Language Models

General advice

<aside> ❗

These are personal opinions. Works well for me. Take them with a grain of salt.

</aside>

• add as much context as possible
• natural conversation as opposed to Google style queries
• use specific keywords to convey context efficiently
• always go through the code/reply critically
• if the topic is new then please be extra careful to fact check concepts
  • it is easier to get productive information on known concepts because we can easily spot hallucinations

Links/resources:

What Is ChatGPT Doing … and Why Does It Work?

ChatGPT Prompt Engineering for Developers - DeepLearning.AI

Personal examples:

R Script

Clickhouse ODBC

Miscellaneous:

https://www.linkedin.com/feed/update/urn:li:activity:7057556541578575872?utm_source=share&utm_medium=member_desktop