The Clearly Podcast

Andy: Today, we're discussing whether ChatGPT is going to take our jobs. This topic has come up because we've seen a lot of posts on LinkedIn and other places about people using ChatGPT for various tasks, including Power BI modeling with mixed results. We want to explore if ChatGPT poses any threat to our jobs now or in the future, and how we should approach it as BI Conn. Tom, can you explain what ChatGPT is?

Tom: Sure. ChatGPT is a type of generative AI that creates text-based outputs by analyzing large amounts of text data from the internet. It's designed to produce language outputs based on user inputs. While it generates text, there are other generative AIs that produce images or sound. When it comes to writing code, ChatGPT uses syntax and keywords to generate something that looks like code, but it might not always function correctly.

Andy: What have we seen in our field, specifically with people using ChatGPT to write DAX code?

Shailan: Some clients use ChatGPT for searches and basic tasks. For example, one client used ChatGPT to create a flow diagram for deciding between different Power Platform apps. The diagram was a good starting point but had some missing elements. It's important to note that while ChatGPT can provide useful outputs, it's not always accurate. It can generate additional questions and sometimes outdated or incorrect information.

Andy: That's a good point. For instance, when I tested ChatGPT with Contoso data, it made errors in the DAX code it generated. If someone doesn't know how to write DAX, they might not spot these mistakes. There's also the issue of data privacy. We should discuss the implications of uploading sensitive data to these models.

Shailan: Exactly. Organizations need to be careful about the data they use with ChatGPT. Sensitive information should not be uploaded. There are policies and regulations, like GDPR, that govern data privacy and sovereignty. Even if you're using demo data, you must ensure it doesn't contain any sensitive information.

Tom: Moreover, you should check if your data privacy impact assessment allows the use of AI tools like ChatGPT. This needs to be clearly defined and agreed upon with data owners.

Andy: Let's shift to the benefits. ChatGPT can be a valuable tool if used correctly. It can help with standard tasks and free up time for more complex work. How can we make ChatGPT our friend?

Tom: We need to practice using ChatGPT effectively, just like any other tool. Learn what it's good at and use it for those tasks. For instance, it can handle standard measures and initial drafts, allowing us to focus on refining and adding more value to our projects.

Shailan: It can help with tasks that are time-consuming but don't require our expertise, like data cleansing or generating initial code. This allows us to focus on strategic work and client engagement.

Andy: Good points. Clients could also benefit from using ChatGPT to interrogate data more effectively. However, we need to ensure data privacy and integrity. In conclusion, while ChatGPT is a powerful tool, it requires careful and knowledgeable use. It's not going away, and if we don't adapt, we'll get left behind.

Shailan: Also, remember that using ChatGPT leaves a footprint. Ensure that you're aware of this and the implications of data privacy.

Andy: Absolutely. And it's important to use the latest version, like ChatGPT-4, which is a paid service. Next week, we'll discuss AI ethics and data handling. Thanks, everyone.

Tom: Thank you. Bye.

Shailan: See you then. Bye.