Data Democratization Will Not Be Visualized

One thing becomes increasingly clear as GenAI (generative AI) and LLMs (Large Language Models) continue to mature: broad adoption of data analytics will not come through data visualization. Instead, it will be a series of natural language prompts. Yes, those prompts may request a visualization, but the entry point and driver of broad adoption of analytics will come through a language prompt, whether typed or spoken.  

Let us put aside the debate regarding bias, accuracy, hallucinations, and other issues with the current generation of GenAI solutions. Those limitations must be acknowledged and mitigated for today’s use cases. However, they recede over time as we a) learn where GenAI does work for us and b) the GenAI systems themselves get better. On the spectrum of correct answers from none/some/most/all, we should all be able to agree that the answer is between “none,” and “all” answers are correct and that it keeps getting better. On top of that, as we work more with GenAI, we learn techniques to improve the accuracy of the answers in how we configure the clients as well as use them. Ethan Mollick introduced me to the concept of the “jagged edge” of GenAI, where some things fall within the sweet spot of what it does well and seemingly similar things fall outside that realm. The fact remains that more and more falls inside the realm of what GenAI can do well.

Three reasons GenAI's model is the future:

Fits my problem

I want an answer to where I should go next.

• Give me the information based on a question

• Provide me with additional information to better inform my decision

• Deliver a pointer on where to look to round out my understanding of an issue

Strategic trend analysis is not an everyday task

• Deeper data analysis is not an “on the spot” activity

• This work will continue and will get used by organizations, but remains specialized

Meets me where I work

GenAI gets embedded in the tools I am using everyday

• GenAI exists in my email, calendar, word processor, spreadsheet, presentation software, etc.

Removes the need to go to a dashboard or set of reports, set filters, and do other exploration streamlines the use of the insights.

• No new tools to learn

• No context switch that disrupts my workflow

• I can get and apply the help I need right inside the software I use for my work

Simple yet powerful

I take the question in my head and ask it the same way I encounter it in my work

• I do not have to map translate question to another system and interface.

No translating my question to a visualization and set of filters to try and suss out an answer

• No more pausing to wonder if a report and the provided filters or underlying data schema can provide my answer

The simplicity of GenAI’s interface and the conversational nature of the interactions gives GenAI an advantage we have not previously seen. This provides a major advantage for any organization that wants a data driven organization but has struggled to reach that goal. Yes, there are limitations with the current generation of GenAI but your organization can absolutely see productivity and work quality improvements by judiciously deploying it at your organization.

‍