Most people would like to have an assistant to help with everyday work tasks and that’s exactly what artificial intelligence (AI) chatbots like ChatGPT promise. They can summarize documents, draft emails, carry out research and much more. As we discussed in the previous article in this series, they are not ‘intelligent’ in the way a human assistant is, so their work needs to be checked for mistakes. Even so, they can save you a lot of time.

One thing to be aware of before using a tool like ChatGPT is that it isn’t like a calculator or spellcheck that will give you the same answer every time. AI chatbots use probability to generate a response that will satisfy a given query, so there is no ‘right’ answer. Repeat a question you asked in a previous session, word for word, and you will get a slightly different response. Change the wording of your question, and the answer could be radically different. With practice, you can use this to explore different approaches to a problem.

In this article we’ll focus on some of the uses for language-processing chatbots. I’ll refer to ChatGPT throughout but there are alternatives, such as Anthropic’s Claude, Google Gemini, and Meta’s LLaMa. Most of these offer a free basic service and charge a subscription for their most powerful AI model or for some other upgrade. Whichever you use, the first six ideas below will be applicable. The last two - the App Store and image generation - are specific to ChatGPT.

1. Writing

The obvious use case for tools like ChatGPT is to have them write for you. Many early criticisms of generative AI came from professional writers, who argued that it produced poor-quality writing. It’s true that the writing is often bland but that misses some significant points.

First, a professional writer isn’t a realistic alternative in many cases. Suppose your job includes writing-up meeting notes to share with colleagues but writing is not your strength. Only a very generous employer would let you hire a writer to handle that, but the low-stakes nature of the content makes it an ideal task for ChatGPT. The writing can be bland so long as it is clear and includes the relevant information.

Second, ChatGPT will give you a solid first draft that you can improve and make your own. That’s ideal if you find the blank page daunting and struggle to get started with writing. But it’s also good if writing isn’t part of your job at all. For example, imagine you run a roofing business and want to publish articles on your website to help promote your business. You might not enjoy writing or may be too busy, and your business is too small to hire a writer. ChatGPT can draft articles on ‘when to get your roof inspected’ or ‘signs your roof needs cleaning’ and you can use your expertise to correct mistakes and add important details.

Finally, you don’t have to use ChatGPT to do your actual writing. You could ask it to help structure your ideas into a logical order before you begin writing. Or you could give it your draft and ask it to suggest improvements. You will find ChatGPT can be a helpful editor.

2. Explanations

Sometimes you need to understand a concept and the internet is your only resource. That’s fine for many simple queries but for something complex it can be hard to find a web page that explains what you need to know and is easy to understand. In cases like that, try asking ChatGPT to explain. Because its database covers so much material it can effectively summarize dozens of web pages, that would take you hours to find, read and understand.

Crucially, the chat interface means that if you don’t understand ChatGPT’s answers then you can ask for clarification - and keep asking until you are satisfied. One note of warning, however. Generative AI is prone to hallucination - or inventing facts - so once you feel like the explanation makes sense, check it against an independent source. Even with this extra step, ChatGPT will still save you time.

3. Generating ideas

Another benefit of the vast amount of training data consumed by ChatGPT is that it’s very good as a brainstorming tool. Ask it for ideas for your next marketing article or even potential new products and it will produce endless lists. Unlike a human, it doesn’t get bored or tired. You’ll reject most of them, but that’s the point of brainstorming. An idea or two might be worth taking further, or something ChatGPT suggests might give you a related idea.

With careful prompts, you can even steer the AI towards specific angles. Ask for ‘overlooked’ ideas, for example, and it should ignore the most common topics. Or look for inspiration in other sectors by asking something like “what can a roofing business learn about marketing from the plumbing sector?” Again, you’ll reject many of the ideas, but the point is to generate as many as possible in a short time - and ChatGPT is ideal for tasks like that.

4. Interview preparation

ChatGPT’s training data contains vast numbers of interviews, articles about interviews, interview tips and more, so it can help you prepare for any kind of interview. You might be interviewing someone for a job, or you might be a candidate – in either case it can suggest questions and possible answers. Or perhaps you are interviewing one of your company’s executives for a marketing profile. ChatGPT can help find questions to ask or avoid, questions that are asked too often and ones that are overlooked. Likewise, if you are the executive who is about to be interviewed, ChatGPT can play the role of interviewer and help you prepare. There’s lots of potential benefit here.

5. Thesaurus

This one is simple but easily overlooked. Search engines do well at suggesting synonyms and there are plenty of thesaurus websites but I’ve found ChatGPT to be more useful because I can ask for specific types of synonym. For example, Merriam-Webster has over 50 synonyms for “leverage” - a common piece of business jargon - but when I ask ChatGPT for synonyms I could use in a business report, I get a much more focused and relevant list of just 10.

6. Summaries

This is a very popular use case because ChatGPT is so good at understanding the context of language. It can summarize the main themes of a document, which is handy if, for example, you want to know whether a report is worth reading in full. You can also get very specific with your queries, asking it to summarize only the parts related to a particular theme, such as supply chain security, or to list only the sections that contain statistics. Essentially this gives you the ability to ‘chat’ with your documents.

7. App Store

Here’s where ChatGPT begins to break the bounds of the large language model. When they first launched, LLMs were very general tools. You could ask them about anything but if you wanted them to specialize then you either had to use a series of prompts to get them to play the relevant role or use a separate service built by developers. Numerous websites sprang up offering the ability to chat with a PDF, for example.

OpenAI, the makers of ChatGPT, realized they could integrate services like these into their own tool as third-party apps. These are known as ‘GPTs’ and, essentially, they optimize ChatGPT for a specific task without you having to write the prompts needed to do it. That allows ChatGPT to be your creative writing coach, say, or your stock market analyst. You can get help with coding, generate diagrams, search scientific papers and much more. The quality of these apps varies, and the App Store is only available to ChatGPT subscribers, but it is worth exploring.

8. Images

We’ve focused entirely on words to this point because that’s what ChatGPT does, but the App Store has extended its capabilities to include DALL-E - OpenAI’s image-generating tool. Once again, this is for subscribers only, but if you sign up then you get access to capabilities that go beyond just DALL-E: there are several custom GPTs to optimise DALL-E for designing logos, creating videos, producing a mood board, or numerous other tasks.

Whatever you plan to use ChatGPT for, the key is to experiment. It only takes a moment to try out an idea, and if it doesn’t produce the result you expected then alter your prompt a little and try again. Or tell the chatbot what you didn’t like about the initial response and ask for one that fits your expectations. Through a process of trial and error you will quickly find out where these tools are a good fit for you and your business.