Delivering return on investment from GenAI [Q&A]
For many people AI has been seen as an answer to all their problems. But a recent Gartner report suggests we could be entering a 'trough of disillusionment' as the biggest challenge when adopting GenAI is estimating and proving the business value.
Getting that value needs creative thinking and integration of GenAI with other technologies that can benefit from its ability to analyze, summarize and create content.
We talked to Saurabh Abhyankar, EVP and chief product officer at MicroStrategy, to learn more.
BN: GenAI has slipped into what Gartner calls the 'trough of disillusionment', is this a fair assessment? What mistakes do you see enterprises making?
SA: The meteoric rise of ChatGPT, Dall-E, and so many other consumer-facing tools over the past couple of years has really driven up the hype around GenAI, and likely the false expectations that it can address every use case all by itself. This is why some analysts are talking about the trough of disillusionment so early in this technology's lifecycle. In the enterprise, where accuracy and reliability are critically important, there are a lot of things GenAI for which is not a good solution for all by itself. For example, on its own, GenAI is not able to reliably analyze structured data or provide perfectly accurate answers to complex calculations using multiple data sources. Enterprise data is extremely complex so you can't just throw AI bots into your Slack channel and expect it to solve data delivery or usability problems.
BN: For those enterprises having success with GenAI, what are the use cases that are working?
SA: Companies that have success with GenAI do three things differently than those who aren't seeing results. First, they pair GenAI with the right tools to address each use case. For example, our customers are combining GenAI with Business Intelligence to help non-technical users gain insights from complex data in a way that is fast, intuitive, and reliable. GenAI doesn't perform the actual data analysis because it's not built for deterministic use cases. Instead, GenAI is incredible at understanding and producing natural language, so it interprets the user’s data request and then translates it for the BI platform, which provides the analysis. GenAI then presents the data in formats easy to understand, including adding narrative explanations of the data with charts and graphs. Simply put, GenAI makes it simple and natural for anyone to interact with a sophisticated BI platform.
Second, companies having success provide clean, organized and governed data to GenAI. Companies with a common semantic layer governing their enterprise analytics data have a head start.
The third thing they do is continually train and enhance their GenAI tooling to learn the language of that organization and the unique needs of their users. The balance between personalization and standardization can be tricky, so enterprises who don’t have the expertise can work with companies like MicroStrategy and our partners to do that job.
BN: There's been a lot of talk about the GenAI hallucination issue. What should enterprises do to address these concerns? What do they need to do to make GenAI more trustworthy?
SA: Hallucinations are a real issue, but there are ways to address it. Reducing hallucinations requires reliable data sources, which is why pairing with trusted BI data makes sense. Additionally, even though GenAI is remarkable at interpreting natural language, people need training on how to ask well-formed, specific questions to get the results they’re looking for. Auto, our GenAI chatbot, is very adept at understanding open-ended questions based on context -- what it learns from the enterprise knowledge base and from its experience with individual users -- but the better the query, the better the results. Strong GenAI implementations will ask for clarification or present a variety of potential interpretations when unsure of what the user is asking for. Finally, GenAI should be able to explain how questions were interpreted and how the answers were derived. Transparency is critical to ensuring accuracy and increasing user trust.
BN: GenAI's flexibility might be one of its biggest benefits. How can it be integrated with other technologies that would benefit from its ability to analyze, summarize and create content?
SA: GenAI is a foundational technology, and over time, it will be woven into and across an enterprise's entire suite of applications and processes. It's the ultimate user interface because it enables non-technical, frontline knowledge workers to interact with data in a natural, intuitive way. For example, integrating GenAI, BI, and ERP data means a supply chain manager could ask, "Create a bar graph showing inventory levels for our key raw materials over the last three quarters," and GenAI will retrieve the data and provide it in the requested format. A sales rep could ask for a summary of a customer's interactions with the company over the past three months, which might use GenAI with BI and a variety of enterprise apps, like Salesforce, ServiceNow, and more. The possibilities are endless.
BN: There are those who believe that GenAI will make data easily available to all employees. If they are correct, what does that mean for the role of the data analyst and the BI industry overall?
SA: We can no longer think about BI as a set of dashboards or a 'destination' that people go to for answers. Those use cases still exist, but to truly become a data-driven organization, data analysts need to consider new ways to deliver information to frontline knowledge workers in a way that is easy to consume and doesn't require extensive training. Ideally, GenAI will one day understand the context of a user's work so it can provide insights proactively, before the user needs to ask for them.
There will always be a need for complex data analysis that enterprise-grade BI can solve, but it's time to broaden the way we think about intelligence, and GenAI is a great catalyst for that. Particularly in the enterprise, GenAI needs the strong, trusted foundation of BI to provide reliable data on which people can confidently base their decisions. As for what this means for the role of the data analyst, they'll be freed up to think more strategically about the application of data for greater efficiency and competitive advantage, because they'll spend much less time building and adjusting dashboards. In the end, GenAI makes BI even more useful to the enterprise by extending its reach from upper management all the way out to the frontline.
Image credit: BiancoBlue/depositphotos.com