“Ask not what AI can do for you, ask what you can do for AI” – so that it works better for you! Here are five ways to get more out of AI chatbots like ChatGPT!

With so many challenges with using generative AI lately, I thought I would mention a few ideas for getting more out of generative AI chatbots like ChatGPT. Plus, I need another day or two for my final post of the year on hyperlinked files and modern attachments!

So, here are five ways to get more out of AI chatbots like ChatGPT! And, yes, some of these were influenced by ChatGPT (specifically GPT-4) because it knows what it wants! 😀

Be Specific in Queries

Clearly define what you’re asking for. Specific questions tend to yield more precise and useful answers. Prompt engineering is the practice of carefully crafting input queries (prompts) to effectively communicate with AI models like ChatGPT to obtain more accurate, relevant, or creative responses. Here are some terrific tips for effective prompt engineering!

Use Specific Instructions and Constraints

ChatGPT can often give you more information than you want, but you can include specific instructions or constraints in your prompt to get a more precise response. For instance, asking for a summary in one to two paragraphs, or requesting an explanation suitable for a particular group or demographic, such as this prompt and response with GPT-4:

Explain electronic discovery to a tenth grader* in one paragraph.

Electronic discovery, often called e-discovery, is like a digital treasure hunt that happens in legal cases. When people or companies go to court, they need to share information that’s relevant to the case. In the past, this meant digging through piles of paper documents, but now, since so much of our information is on computers and online, it involves finding and sorting through electronic data. This can include emails, texts, files on your computer, social media posts, and more. The goal is to find and collect this digital information in a way that’s organized and fair, so that it can be used as evidence in court. Think of it as gathering all the important digital clues to help solve a legal puzzle.

Take an Iterative Approach

Models need training to get things right – sometimes the model just misses the mark the first time or two on the information you’re trying to get. Use the responses you get to refine your prompt. If the initial answer isn’t quite what you wanted, tweak your prompt based on the response and try again.

Use Feedback Mechanisms

Have you noticed that every interaction with ChatGPT gives you “thumbs-up” and “thumbs-down” icons to tell how useful the responses are? If the platform allows it, provide feedback on responses. This can help improve the accuracy and relevance of future responses (assuming you don’t mind that it uses that data for training, of course).

Stay Informed about Updates

Keep up-to-date with new features, improvements, and usage policies, as generative AI technologies are rapidly evolving. Just two months ago, ChatGPT didn’t include image and voice capabilities, Browse with Bing & citations, analysis of PDFs or the ability to create your own GPT (OK, that one may still need some work). This technically doesn’t evolve – it morphs and mutates! 😮

So, what do you think of my five ways to get more out of AI chatbots like ChatGPT? Please share any comments (including other ideas!) you might have or if you’d like to know more about a particular topic.

*Why a tenth grader? Because I have one! 😉

Disclaimer: The views represented herein are exclusively the views of the authors and speakers themselves, and do not necessarily represent the views held by my employer, my partners or my clients. eDiscovery Today is made available solely for educational purposes to provide general information about general eDiscovery principles and not to provide specific legal advice applicable to any particular circumstance. eDiscovery Today should not be used as a substitute for competent legal advice from a lawyer you have retained and who has agreed to represent you.