My week with AI
How Copilot, Microsoft’s AI chat service, boosted my productivity and efficiency as a BI consultant.
AI is transforming the way we work with data. As a data enthusiast, I was curious about how AI could enhance my work, so I started using Copilot, formerly known as Bing Chat Enterprise. Copilot is a Microsoft service that uses AI to perform various natural language tasks related to writing and creating, such as generating text and images, summarising documents, answering questions and more, in a private and secure environment. Copilot does not store or train on organisational data, and it uses encryption and authentication to protect your data in transit and at rest, so you can use it with confidence knowing your privacy and data are protected.
In this blog post, I will share my initial experiences using Copilot to transform my writing process, generate and analyse code, and create images with AI based on text prompts.
Transforming my writing process
As a busy professional, I’m always looking for ways to save time and automate some of the repetitive aspects of writing. I first tried Copilot when I needed to quickly summarise some key points from a collection of notes. It took me a few tries to get the right prompt, but once I did, I was amazed by how fast it was to create a two-page document from just a bullet-point list. Since then, I’ve used Copilot for various writing tasks, such as crafting introductions and summaries, adjusting the tone and style of my text, creating documentation, and even helping with writing blogs.
The most valuable feature of Bing chat for me so far is its ability to restructure my sentences to enhance their clarity and coherence, saving me the time and hassle of searching for the perfect words and speeding up my editing and proofreading process. This has freed up my time and energy to focus on more innovative and challenging tasks.
If you want to try Copilot yourself, here’s a tip: be specific about what you want. You can ask Bing to rewrite your text in different tones, such as more professional, more concise, more impactful, or more constructive. The secret is in the prompt! If the first response isn’t quite hitting the mark, ask Copilot to rewrite it with a different tone. Consider the words that capture the user experience you aim for when people read your text and include them in your input.
Generating and analysing code:
As a BI developer, I work with various types of code daily, such as SQL queries, Dax, PowerQuery, and Power Fx. Copilot is designed to work as an AI pair programmer tool that can handle dozens of languages and frameworks and learns from billions of lines of public code. I decided to test Copilot's capabilities in this area and see how good it is at comprehending, debugging, and creating code. These are some of the features that I experimented with:
- Understanding code: For its supported languages, Copilot can simplify the meaning of most code snippets with natural language. To get Copilot’s explanation of the code, just paste the code as the input and ask what the code does. I found this feature extremely helpful for reviewing code that lacks comments and for creating clear documentation for non-technical users. It is also a great tool for when you are learning new coding skills or concepts. Tip: you can also ask Copilot to rewrite the code with comments added, or even to reformat the code for better readability.
- Debugging code: Copilot can be used to find and fix bugs and errors in your code. Of course, it’s not perfect, and you need to provide some context about how the code is used, but I found it helpful for spotting syntax errors that are easy to overlook with the human eye. It can also search and interpret a vast amount of documentation, saving you time and effort scrolling the internet for answers. Next time you find yourself frustrated with code that doesn’t work, I’d recommend copying it into Copilot for support and saving yourself the headache.
- Code generation: Based on your natural language input, Copilot can create code for various use cases and coding languages. However, I found that it is not always reliable in the quality of code produced, as it is often not optimally written. This might not always be clear unless you have decent experience and knowledge of the language. However, you can improve the code generated by asking Copilot to rewrite it in the most efficient way possible. I would say that is a good feature for overcoming coding knowledge barriers, but it should be used and reviewed with caution.
Image generation:
One of the most astounding features of AI is the ability to generate images from text prompts. I discovered this feature when researching for this article, whilst Copilot was still Bing Chat Enterprise, and I decided to use it to create an image for this blog post. I entered a text prompt describing image requirements and the colour palette to use. Within minutes four different images based on my prompt were produced for me to choose the one I liked most. I also had the option to further customise the images by adding more details to the prompt. You can see the results via the images used in this post, or by watching the recording at the bottom of this article to see how I did it.
I was blown away by this feature, even though I have less of a need to use it frequently in my day-to-day role. It shows how far technology has come and how quickly since the first attempts at AI image generation.
Final thoughts…
I have found that the key to using Copilot is to craft good prompts that elicit the best responses for your needs. This may take some trial and error, but it is worth it. I love how user-friendly Copilot is — no coding knowledge is required, just natural language. This makes it accessible to people in a wide variety of roles, not just technical.
There is an element of using AI to achieve fast results that feels on some level like a shortcut that cheats the system, and we risk losing some of the skills that we become to depend on AI for as time goes by. However, it also gives us the chance to acquire new skills in AI technologies, and the improved efficiency of using AI tools like Copilot will likely result in cost benefits for organisations and possibly higher job satisfaction for employees who can avoid tedious tasks.
Ethical issues of AI are also important to consider, such as disclosing our use of AI and obtaining permission to use the data. Finding the balance between using AI ethically and knowing its limits will be one of the difficulties we face as we rely on it more.
I will continue to use Copilot and I am eager to try out more AI tools. Please let me know how you have used AI to improve your work and what you recommend me to check out next. Also, don’t forget to follow me on Medium to stay updated.
About the Author
Sarah King is a Data Visualisation Consultant here at Version 1.
