ChatGPT: A marketer’s guide

Knowing ChatGPT's strengths and weaknesses can help you decide where and how to use the tool for marketing campaigns.

Chat with MarTechBot

Artificial intelligence has dominated headlines recently as the world began to play with a new tool: Chat Generative Pretrained Transformer, better known as ChatGPT. 

The tool quickly attracted techies and non-techies because one prompt can generate a response usable by editors, PR teams, developers or executives to create white papers, software programs, client presentations, press releases and more.

As marketing professionals, it’s critical to learn what it can and cannot do today, how best to use it for our campaigns — and what’s in store for us.

What ChatGPT can do today

Launched by OpenAI in November, ChatGPT is built on top of OpenAI’s GPT-3 family of large language models (LLM) and enables interaction with a model via a conversational user interface. 

ChatGPT computes the next most probable set of letters or words when given an initial starting phrase or “prompt.” It was trained on 300 billion words taken from books, online texts, Wikipedia articles and code libraries — reportedly using a snapshot of the internet as of 2021. 

“As a language model, ChatGPT is best used for tasks that involve natural language processing, such as text generation, text completion  and answering questions. Specifically, ChatGPT can be used for a variety of tasks including:

1. Chatbots: ChatGPT can be used to power chatbots and virtual assistants that can engage in natural language conversations with users.
2. Content creation: ChatGPT can generate human-like text on a variety of topics, which can be useful for content creation, such as writing articles or generating product descriptions.
3. Question-answering: ChatGPT can provide answers to questions posed by users, such as providing information about a particular topic or field.
4. Translation: ChatGPT can be used to translate text from one language to another.
5. Text summarization: ChatGPT can be used to summarize longer pieces of text into shorter, more easily digestible summaries.
6. Personalization: ChatGPT can be trained on specific datasets to provide personalized responses or recommendations based on user behavior or preferences.

Overall, ChatGPT is a versatile tool that can be used in a wide range of applications that involve natural language processing.” 

This section was generated by ChatGPT with the prompt: What is ChatGPT best used for?

Today, the outputs of ChatGPT are impressive in documentation areas. It can assist — but not replace — humans in writing reports, outlines, creating press releases, books or developing surveys. It helps writers and editors get started in developing pieces of content. When you have writer’s block and are stuck, ChatGPT can come to the rescue.

Dig deeper: When I asked ChatGPT to write an article about ChatGPT

What are ChatGPT’s limitations? 

As the log-in page of ChatGPT points out, it does have limitations, like any new technology. ChatGPT may occasionally generate incorrect information, including harmful instructions or biased content.

Remember, this tool is based on a snapshot of the internet in 2021, a year of pandemic mania, racism and unmitigated lying. The tool is only as good as the data that supports it. That’s why it cannot answer questions about events that happened from 2022 to date.

Below are some other areas where ChatGPT has limits:

At best, it is a complement to search and only a partial alternative. ChatGPT focuses on generative approaches to answers rather than artifact discovery (e.g., a particular document or sentence). It also does not tell you from what source it pulls the information, meaning anything it reproduces is not attributed to the original author. 

Language translation

ChatGPT can translate simple sentences into traditional languages. However, it can have problems with translations for domain-specific areas and languages well beyond the Mediterranean languages. A human translator should be engaged to validate the translation.

Privacy or confidentiality 

ChatGPT users should treat the information generated as a public site post and avoid publishing personally identifiable, company or client information. Users’ conversations with ChatGPT could be used for training new models and will be reviewed by trainers. You can’t delete specific prompts, so be careful what you share. While you can delete an account, this will not delete the training data.

Accuracy of output

Users should carefully evaluate the inputs and outputs of the tool for misrepresentations and biases. ChatGPT is enhanced to align with the trainers’ preferences rather than verified facts. This means that output is plausible but not reliable for many use cases. Moreover, bias might be present in the large datasets that train the model. 

Here’s an example close to home. I used ChatGPT to generate a statement about my accomplishments and background. For the most part, it was very accurate, but when it had me starting my career as a software programmer, I immediately saw the tool’s limitations. Since I was in the technology business and had worked for several large tech companies, it inferred that I must have started as a software engineer. I didn’t. I started in marketing.

Dig deeper: Three things ChatGPT needs which only you can provide

How marketers can best use it

Knowing ChatGPT’s strengths and weaknesses can help you decide where to use the tool. Since natural language processing (NLP) is what it does best, start with the tasks that you might need as a marketer.

  • Writing point-of-view papers.
  • Developing website content.
  • Updating social posts.
  • Creating blogs.
  • Writing bylined articles.
  • Generating press releases.
  • Writing program or project documentation.
  • Developing a marketing strategy.
  • Drafting legal agreements (first draft of your requirements to give to your legal department).

Usually, half of your time developing some of these pieces is spent researching. ChatGPT will be a good go-to for research assistance if you remember the limitations above. 

Here are two last pieces of advice when using ChatGPT:

  • Develop a prompt that will narrow the output for the tool. I’ve found it works best when the topic is specific and ask it to simplify the response.
  • Think it will save you time? You may be disappointed. It is best used as a “hole” filler if you are limited on staff or researchers. But if you have a professional writing staff, then ChatGPT should be viewed as any other tool they would use to help refine their craft. View it in the same category as spellcheckers or Grammarly. 

What marketers can expect in the future

ChatGPT belongs to the category of growing generative AI tools, where you can also find Dall-E, which generates digital images from prompts. In February, Google announced it would enter the market with Bard, its version of ChatGPT. From what we know of the two tools, here are the key differences: 

ChatGPT (OpenAI and Bing)Bard (Google)
• Purpose: to predict the next word
• Can be used to:
— Generate sentences
— Summarize
• No Open Source Code
• Cannot be re-trained
• Purpose: to produce abstract expression
• Can be used for:
— Search
— FAQs
— Translation
• Open Source Code
• Can be re-trained with your data 

As this tool category expands, expect better outputs and current references. (Google will use its own data set to train Bard). The critical question is: “Will it replace marketing writers?”

No. Like all tools, it will supplement and complement the writers’ capabilities but not replace them. After all, there is a limit to what a computer can create.

Sentient computers are still several decades away. In the interim, we need to get used to using tools like ChatGPT because the limitations of technology will always be in the hands of humans.

Dig deeper: Does ChatGPT pose an existential threat to marketers?

Get MarTech! Daily. Free. In your inbox.


Opinions expressed in this article are those of the guest author and not necessarily MarTech. Staff authors are listed here.


About the author

Theresa Kushner
Contributor
Theresa Kushner is passionate about data analysis and how it gets applied to today’s business challenges. For over 25 years she has led companies – like IBM, Cisco Systems, VMware, Dell/EMC – in recognizing, managing, and using the information or data that has exploded exponentially. Using her expertise in journalism, she co-authored two books on data and its use in business: Managing Your Business Data: From Chaos to Confidence (with Maria Villar) and B2B Data-Driven Marketing: Sources, Uses, Results (with Ruth Stevens). Today, as the Data and Analytics practice lead for NTT DATA, Theresa continues to help companies – and their marketing departments -- gain value from data and information.

Get the must-read newsletter for marketers.