The power of AI in data storytelling
Image created by GTP-4

The power of AI in data storytelling

In a rapidly changing data landscape, leveraging artificial intelligence (AI) and generative AI can significantly enhance our approach to data storytelling. This newsletter is the third installment in my AI and data storytelling series and delves into how these technologies can help answer the final question of data storytelling: What actions should we take based on the data we analyse?

In the second newsletter of this series, I talked about using AI to help uncover trends and insights in the data, which is the first data storytelling question. However the focus this time will be on the action. When working out how to respond to trends and insights, its's important to comprehend the underlying reasons behind data trends and insights. Often, when looking at data our minds can fall into the trap of confirmation bias, leading us to jump to conclusions that support our existing beliefs. This cognitive bias can hinder our ability to see the broader picture and consider alternative explanations, and therefore alternative responses.

Daniel Kahneman's book "Thinking, Fast and Slow," distinguishes between two types of thinking: System 1, which is fast and automatic and System 2, which is slow and deliberate. Most decisions we make fall into the System 1 category, which can lead to snap judgements about data trends. AI and generative AI can help counteract this bias by providing alternative explanations for data phenomena, encouraging us to think more critically before arriving at conclusions. Sometimes it can be as simple as asking generative AI to provide reasons for some of the trends you see. They may be way off, but at least it forces your brain to consider other possibilities.

As well as confirmation bias, another common challenge in data analysis is action bias. This occurs when individuals, upon identifying a trend, rush to implement the first solution that comes to mind, often neglecting a more thorough analysis of options. AI can facilitate a more structured thought process, suggesting possible actions to address identified issues and encouraging teams to explore a variety of potential responses before committing to a course of action.

Recently I was asked to do an analysis of some sales data where we noticed a big slump in sales. Using AI tools, I was able to quickly come up with a few possible actions. However, in light of the current economic climate in Australia, marked by rising living costs and reduced discretionary spending, actions that relied on the premise being to 'increase sales' were clearly not suitable. It's here that generative AI comes into its own because you can then narrow down the prompt to generate targeted solutions that accounted for these specific economic conditions, and were framed in this particular context. While some suggestions echoed our own brainstorming, the AI provided fresh perspectives that prompted deeper consideration of the next steps.

AI also has the potential to assist in creating engaging data stories. While it may lack the empathy and contextual understanding that human storytellers possess, generative AI can efficiently produce initial drafts, summarising complex datasets and identifying key trends. This serves as a valuable starting point for refining narratives and deepening insights. While AI-generated outputs should definitely be seen as first drafts, they can significantly reduce the time and effort involved in crafting data stories. By using AI to handle preliminary summaries, data professionals can focus their energy on refining and enhancing the narrative.

Integrating AI into data storytelling reveals the technology's vast potential. By understanding the "why" behind trends, leveraging AI to suggest actionable steps, overcoming action bias, and utilising AI for narrative creation, data professionals can enhance their decision-making processes. As AI continues to evolve, the landscape of data storytelling will undoubtedly transform and I, for one, am so excited about it.


This newsletter started as a podcast; if you'd like to listen to this episode and/or follow the podcast, check it out here.

______________________________________________________________

I am a data storyteller and grounded researcher, and I help organisations use their data more effectively and help them tell great data stories. If you'd like a hand with data storytelling or strategy, I'd love to chat with you.

Keynote speaker | Author | Facilitator

Eshwaran Venkat ?

Empowering Restaurants with Market Insights ? Co-Founder & CTO at Dotlas ? UC Berkeley MIDS ? Sharing Industry Trends for Retailers

1 个月

While this could be very helpful to check for human biases as you correctly put, it's also going to be difficult to manage hallucinations. I do think that LLMs have a huge place in interpretation and recommendations from descriptive stats.

Winston S.

Senior Info/Data Management Professional - Experienced Senior Leader in multiple Data Management disciplines - Data Strategy | Data Governance | Data Protection | Data Privacy

1 个月

Using AI to provide additional reasoning to the already performed data analysis, demonstrates why data-driven is not always the right approach as information is being sourced externally to enrich the context and information presented. The influence of bias and hallucinations, and the lack of vetting any sources the AI model will have used may lead to a decrease in reliance on the analysis presented. It's also required that any findings presented, regardless of source, is socialised with the requestor to ensure it is presentable to a wider audience. We still live in a society where the "unbiased, this is what the data says" approach is not fully adopted, people still want data to tell a particular story that fits close enough to the narrative in their head. Often the report is a supporting artefact to a narrative that is being built, this must be appreciated.

要查看或添加评论,请登录

社区洞察

其他会员也浏览了