Is AI Making You a Better Business Analyst?
Not so fast
No, I don’t have a course on the best ChatGPT prompts for business analysts. I’d rather teach how to create helpful diagrams.
I don’t talk about how using AI makes me a better business analyst. Because it doesn’t.
First of all, “using AI” in this context just means “using an AI chatbot”. A chatbot is a Large Language Model (LLM), one of a myriad AI algorithms. Chatbots help us churn out content (text) in response to our prompts. So a chatbot may increase the speed of your research and make your writing more fluent, but will it improve the quality of your analysis?
More than that, will using AI chatbots allow you to grow and learn new BA skills? Or does it make you think that there is no point in learning the skills because “we now have AI”? There is a term for this — “cognitive offloading”. It can’t be good for us.
Studies have already shown that AI can be harmful to people who turn to it for mental health advice because of its sycophantic tendencies.
And don’t even get me started on all the money-making enterprises that produce “AI-certified” specialists who do not understand the difference between a stochastic and a deterministic algorithm.
AI chatbots make a lot of logical mistakes and can produce “confident-sounding slop” faster than you think.
Generative AI is very good at creating an illusion of quality output, until you dig deeper.
But will it help you think? Will it show you the gaps in your thinking and the problems in your analysis?
Chatbots are good at producing the most likely output, “converging to the mean”. They’ll give you the middle, the most obvious. The exceptions, the weird stuff, the unconventional scenarios are a different story. Yet, this is where a BA brings value, by flushing out the gaps and inconsistencies, rather than by stating the obvious.
I see my students relying on ChatGPT as a crutch in their assignments. The paragraphs all come out in the same format and of a consistent length. But if I ask a student to explain what a particular sentence means, they just look at me. They have no idea.
Yes, AI can summarize your meeting for you. It may be helpful. But it sometimes misses critical points. If I’m not taking my own notes in parallel, I may discover that a vital conclusion is completely missing in the AI transcript, vanished without a trace.
Yes, AI can create action items from a transcript. But they are so wordy that it takes me 10 minutes just to read them and then shorten them to something that works for a normal person’s attention span. Most of the time, I would make 2–3 bullet points during the meeting and use these instead. Besides, when we write, our brain processes this information better, and we understand and remember it. When I read long-winded AI-generated sentences, they sound like many other sentences before them and all merge into one droning voice. I don’t remember much.
I’ve seen business analysts go to all kinds of trouble to write scripts and instructions for an AI tool that would generate a diagram. The diagram would look like a long and tedious sausage that I would not even be interested in editing. Sometimes, it’s more harmful than helpful.
Yet, with practice, a BA can create a diagram from the same prompts in 5-10 minutes. The information would already have been collected, since they need to feed it to the AI tool as inputs. And while creating a diagram, their brain will process the information again and may detect gaps and new questions, arising from trying to make the diagram fit.
Perhaps the tools will get a bit better, but at what cost? I don’t only mean the energy and water. The other cost is when we start to delegate thinking to these tools that give us an illusion of competence and excellence.
Now, don’t get me wrong, you can get efficiencies from using AI tools. But we are focusing on the wrong things here.
We should talk more about how to develop enterprise-scale AI solutions, contributing as business analysts to these efforts.
We should talk about business analysts gaining more data skills, data analysis techniques, understanding data structures, how data ingestion and data preparation works.
We have to emphasize that AI algorithms are stochastic, not deterministic, so they will make incorrect predictions by design, and business analysts need to think about requirements for handling edge cases, mitigating, and reducing the risks of errors and potential harm from AI mistakes and hallucinations.
There is new research shows that AI chatbots are becoming dumber when fed brain rot from social media.
Independent thinking, critical analysis, and the ability to see outside your echo chamber is becoming rarer. However, this critical thinking is essential for business analysis and for solving business problems.
So take a piece of paper and draw a diagram by hand, thinking through every shape and every line. This is a good exercise to keep your brain sharp.
I teach business analysis and data analysis fundamentals, BA tools and techniques, how to run analytics projects and define requirements for data science and AI initiatives.
Website | Contact | Speaking | Training | Mentorship library | Courses | Books | Downloads | LinkedIn | YouTube | BlueSky
