Feature
ChatGPT had an estimated 100 million users within two months of its rollout. Six months later, ChatGPT users have dropped by nearly 10%, according to Internet data firm Similarweb.
The dropoff, it turns out, isn’t because the tool isn’t delivering on its promise to increase efficiency in the workplace. And it’s not because companies are putting boundaries on how employees can use AI. In fact, a recent survey of 3,000 managers by software company Beautiful.ai found that 93% of managers encourage their employees to use AI tools to improve their work.
No, the reason for the declining user count was caused by how employees are using the technology.
Getting Around the Prompts
If you’ve used ChatGPT, you know that the usefulness of generative AI in the workplace depends on the prompt.
“Asking ChatGPT for information isn’t efficient because you have to fact-check it,” said Emily DeJeu, assistant teaching professor in business management communication at the Carnegie Mellon Tepper School of Business in Pittsburgh. But asking the tool to assist in completing a specific task such as writing a memo, fixing grammatical errors or analyzing data can save employees time, she said.
The key is identifying the most valuable use cases for your business and processes.
Julie Rutherford, senior director of marketing at TorchLight Hire, encourages employees to use ChatGPT for tasks like writing outlines for blog posts, suggesting questions for webinars and drafting social media posts. “We often have to rewrite them and add our own flare because they feel generic, but it can really jumpstart your work,” she said.
Related Article: Generative AI Writing Job Descriptions: Adult Supervision Required
5 Content-Driven Use Cases for Generative AI
Here are five ways managers can encourage employees to use ChatGPT to increase productivity.
Write a First Draft
This is perhaps one of the most known use cases of generative AI. Tools like ChatGPT can help employees get around the infamous writer's block by providing them with an outline or a rough first draft.
If, for instance, an employee is working on a presentation on why a potential client should hire your firm, the person can ask ChatGPT the information that should be included to make the case. “You’re not relying on it to make the argument for you but to help you know what information to include in the presentation to make your argument,” DeJeu said.
An employee also could ask ChatGPT to write an email to a client about an upcoming meeting, and then ask ChatGPT to write it again in a slightly different way. The employee can then compare the drafts, pulling language from each version, adjusting the tone of the email to fit its intended audience and then add factual information, DeJeu said.
“Your judgment is critical for the final product,” she said, but it’s a very efficient way to get started.
Fix Grammatical Errors
Once a document has the correct tone and information, generative AI can help fix any grammatical or punctuation errors before publishing or sending it.
Even better, it can also explain the rationale behind the edits, DeJeu said. This can be especially helpful for someone who is writing in a language that isn’t their native language. “As business is becoming increasingly global, a team in the U.S. might have to communicate with a team in South Korea, and generative AI can help with correcting errors in translations,” she said.
DeJeu does warn that generative AI is not as reliable when it comes to improving the wording or tone of a document. “If you were to prompt it to make a document sound better,” she said, “it will take words out, combine sentences and possibly make changes that aren’t as useful.”
Perform Repetitive Tasks
Generative AI can be of tremendous value with repetitive tasks such as writing short code segments and debug code, said Andrew A. Li, assistant professor of operations research at the Carnegie Mellon Tepper School of Business.
It can also suggest the best programming package to use to write code. For instance, Li said, an employee could prompt ChatGPT by typing, “I would like to program an app that sorts photographs, what existing programming package is best suited to help me.”
An employee could also feed code that isn’t working properly into ChatGPT and ask it to debug the code. “Any person programming should be encouraged to play around with ChatGPT the same way writers played around with the Thesaurus function on a computer when it was first introduced,” Li said. Any interaction with ChatGPT should of course come with the assumption that any information put into the system will be used to train future iterations, he added, so employees must exercise caution and judgment on what they enter into the system.
Related Article: Better Together: How Creatives Can Benefit From Generative AI
Generate Graphs and Maps
Companies that subscribe to ChatGPT Plus have access to Code Interpreter, which allows ChatGPT to analyze data and create charts and maps using raw data from an Excel document.
“Everyone can now be a data scientist,” said Clayton Durant, director of emerging media and platform strategy at MikeWorldWide, a full-service public relations agency in New York City.
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
Durant, who is leading the company’s AI adoption strategy, has been teaching employees to use Code Interpreter to build graphs and other data visualizations. For instance, the company recently used Code Interpreter to build a historical look at the search term “Barbie,” using more than 600 Excel cells.
“Instead of taking hours to create the graph, an employee could feed the Excel document into Code Interpreter and build the asset in about 10 minutes,” he said.
Companies can even go a step further and use that capability to build original social media content.
Boost Creativity
Considering the risks and potential hallucinations of using generative AI at this point in its development, employees can’t fully rely on the tool’s output without first vetting it, but managers should encourage them to view the technology as a collaborator or coworker, Li said.
In fact, a recent study by MIT Sloan Management Review found that 60% of respondents view AI tools like a coworker.
Ask ChatGPT the same questions that you’re asking your colleagues, DeJeu said. “As long as you’re not relying on generative AI tools for information that needs to be true, it can provide creative ideas,” she said.
For instance, employees can ask for input on how to perform a task, for ideas on specific topics or suggestions on how to reach a specific goal.
“It can really help employees get started on tasks that seem really big,” Rutherford said. It can create an outline or raw content that the employee can then start iterating. “It’s important for managers to encourage that type of collaboration and help employees to understand that using generative AI will free them to use their brains more, not take their jobs.”
