If you’ve had a less-than-stellar experience with ChatGPT, you may have chalked it up to “Garbage in / Garbage Out.” 

But this is wrong as it views the “data” that ChatGPT has access to as the valuable part, so when that data is “wrong” it’s easy to write it off.

The data in ChatGPT is not the valuable part.

The language processing capabilities - trained on that data to understand how language works, not the contents therein  - is where the magic lies.

ChatGPT without Context is a Demo

Think of it this way: when you run a ChatGPT prompt without providing additional context, you really just get a demonstration of what it’s capable of.

Barring any additional context, ChatGPT will fulfill your request by using the information it was trained on to show you what it can do. That’s all.

Remember that. It’s a demo. And, yeah, sometimes the demo is good enough.

But often it’s not.

So, if you've had a less-than-impressive experience with ChatGPT, it's important to understand that running a prompt without providing additional context probably isn’t going to get you what you want.

Don't Blame the Prompt, Blame the Context

To truly unlock the power of ChatGPT, you need to provide it with training data that aligns with the task you're asking it to complete.

A lot of people spend a lot of time trying to craft the most impressive prompt, failing to realize that it’s not the prompt, but the training data and context. Fix those issues, and your prompts will work so much better.

Two Types of Training Data

In this course, for example, we talk about Task context and Session context.

Task vs. Session Context

Task context is context about the task (or job, or directive) you’re trying to complete and is not session-specific. This can include task objective, domain-specific information, formatting directives, etc. 

In a Customer Success, Marketing, or Sales setting, this Task context would be centralized so everyone on the team is generating consistent output.

But then you need to have context that’s specific to that task being completed in this session.

Case Study Example

For example, if you're a Customer Success Enablement specialist or in Marketing and you need to create a Case Study, you’ll need to gather Task context (how to structure the output: Style Guide, Template, etc.), as well as Session context: Customer info, goals and progress, results, quotes, etc.

Task context is consistent from session to session, but each session has specific context, in the example of a case study, it would be context about the customer and their results.

ChatGPT is a Language Model, not a Search Engine

Just remember, during the training process, ChatGPT was fed massive amounts of text data from the internet and learned to recognize patterns and relationships between words and phrases. 

This allows it to generate human-like responses based on the input it receives. But just like humans, if it doesn't have the right context, it can provide inaccurate or irrelevant output.

So, it's important to understand that ChatGPT is not a search engine.

It's a powerful language processing system that can help you write engaging emails, produce knowledge base articles, generate case studies, and much more. 

But to do so effectively, you need to provide it with the right context.

Training ChatGPT is Critical

If you don't train ChatGPT correctly, you'll end up with emails that are too long and contain cliches and tropes that'll cause the recipient to ignore you, knowledge base articles about features not present in your product, and case studies about fictional customers and their fake results.

ChatGPT can be an incredibly valuable assisting technology for your team.

But you have to use it right, and that starts with understanding what it is and where its real value lies.