Data visualization tools, desktop data discovery tools, and visual analytics are examples of traditional self-service BI tools that business analysts embrace because they provide a user-friendly way of quickly turning data into insights. These tools are geared toward business analysts that have the skills and knowledge to acquire the right data sets, perform the analysis, and present the insights needed to solve a business problem.
Often, these data sets acquired by business analysts are not governed or managed by IT, but this is acceptable because business analysts have enough business knowledge to evaluate whether insights are reasonably accurate to address the business problem. Business analysts also have the skills to best present analysis in the form of beautiful charts and reports to make it easy for others in the business to interpret insights quickly for decision making. In other words, the art of insight commonly outweighs 100% accuracy or governance.
Attend Oct 10 webinar ‘Governance in the Age of AI’ with Brad Peters, Infor’s SVP and GM of Analytics and BI. Register now.
The era of AI-enabled BI changes everything. Machine-generated insights can remove business analysts entirely from the analytic process. However, applying AI to ungoverned data sets without human oversight creates a risk of inaccuracy and unreliability in the insights delivered to the business user or decision-maker.
Fortunately, governed data discovery tools do exist that provide the unique capability of self-service BI to IT governed data sets. Some of these started life as enterprise reporting tools and expanded to self-service BI. Others started out as self-service data visualization tools and remain ungoverned. Some, like Birst, were built from the ground up for governed data discovery to offer a balance between IT control and business analyst freedom. With Birst, integrating AI in the analytic process ensures more reliable machine-generated insights.
I spoke with Brad Peters, Infor’s SVP and GM of Analytics and BI, to uncover why governance is even more critical to BI in the age of AI.
1. In the world of BI today, how will AI solve major challenges such as getting more users to use data and analytics for decision making?
AI brings the skillset of the expert analyst and puts that into the machine. The expert analyst brings two key characteristics for creating insights: 1) knowledge of the business and the ability to 2) analyze data.
A governed semantic layer puts the knowledge of the business analyst into the BI tool and makes it broadly available in a way that desktop discovery tools don’t. And then AI takes the skill of analyzing data and finding insights and makes that broadly available, as well. The AI-enabled BI system processes data using the guidance of the governed semantic layer to deliver reliable, contextual analysis with no human involvement. Previously, this process was only in the head of the business analyst.
The governed semantic layer is critical in encoding business knowledge about the relationships between data and how data makes sense to be viewed. Without the semantic layer, it’s just a bunch of data. The semantic layer brings meaning to the data and provides a way of expressing that to the machines. AI can then browse data using that knowledge of semantics and can uncover novel and interesting discoveries as an analyst would, but AI-enabled BI generates these insights on behalf of people who are not expert analysts.
Be aware that many desktop discovery tool vendors claim they deliver a governed semantic layer, but in reality, they don’t offer a single, shared semantic layer. When these tools are deployed in a large organization, desktop users in each department end up creating their own definitions with no ability or architecture to reuse the semantics or enforce a single version of the truth.
2. We have seen a rise in BI user adoption with the era of self-service BI and large-scale computer power for better response times. What are the risks of applying AI to desktop discovery tools?
First, I want to point out that although desktop discovery tools are very popular, they do require an extensive skillset to gain valuable insights. The average business analyst is not going to be able to do much with the tool and will end up using it as a reporting engine. That is why a governed semantic layer in an enterprise BI platform is so important because it takes that expert knowledge and makes it broadly available. And when AI gets a hold of it, it scales these expert analysts.
The risks of applying AI to desktop discovery tools that don’t offer a robust set of semantics is that AI is fundamentally limited in what it can figure out, resulting in the garbage in, garbage out. Desktop discovery tools are unguided and rely strictly on expert analysts to bring meaning to the data. Analysts can sift through the garbage and iterate, whereas a robot can’t without rich semantics.
3. What will it take for AI to conquer reliable report creation?
Composition in the semantics so that AI can understand what’s important. What an analyst brings to the table is an understanding of significance and importance and broader business relationships outside of just data relationships. If the semantics can bring that to the data, then AI has some leverage. Otherwise, it runs the risk of evaluating everything as the same, importance, and that’s not terribly useful.
4. In theory, AI can completely automate the analytic process. In reality, what BI tasks are suitable for automation and why?
Anomaly detection is a big area since AI is good at quantifying what is normal to measure non-normality. Much of BI has historically been about anomaly detection and monitoring the business to know if something is going wrong, for example, fraud detection.
Discovery is another area. AI is useful for identifying trends and patterns that may be an analyst didn’t see. Segmentation, clusters, and analysis where you’re trying to understand if there is a critical mass of groups that display similar sets of characteristics so that you can group them together and treat them separately from other types of groups. Those are all things that AI is good at.
Prediction is an area people want to use AI for; I think that’s very much in its infancy. To make excellent predictions, it requires exceptionally good semantics and very tailored prediction techniques.
5. Self-service BI enables business analysts to create artistic stories that help executives and decision-makers easily understand data. How satisfied do you think these business users will be with machine-generated insights versus analyst-generated?
That’s a classic problem. Also, that’s one of the areas where Birst has some of the longest experience in this industry, going back 15 years, providing machine-generated insights to normal business users. There is also a cultural barrier that needs to be addressed. People inherently don’t trust machines, and there’s a cognitive bias when users know something is machine-generated. They tend to focus on why the machine is wrong as opposed to trying to understand why it’s right.
The challenge is figuring out how to explain machine-generated insights, so people understand why the insights were generated. The more sophisticated the algorithm, the less explainable it is. That’s an ongoing area of work for all vendors.
To hear more of my interview with Brad Peters, register the Oct 10 webinar ‘Governance in the Age of AI’
Mona Patel works in Birst’s Product Strategy team. With more than 20 years of experience building analytic solutions at The Department of Water and Power, Air Touch Communications, Oracle, MicroStrategy, EMC and IBM, Mona is now growing her career at Birst. Mona received her Bachelor of Science degree in Electrical Engineering from UCLA.