AI For The Sake Of AI? For Sure Not – Says The BA!
By Elizabeth Larson
Many organizations–around 72% according to Harvard Business Review– are finding that their AI initiatives are not meeting expectations[i] There are many reasons for disappointing AI results, among them that many organizations:
- Chase fascinating new technology with no clear vision of the business proposition. In an article entitled How to Build a Business Case for AI[ii], Gartner warns that AI projects are vastly different from other projects, and states emphatically that “CIOs must put forward a solid business case.” Forbes states that a business case is necessary, and the organization must look beyond just ROI (return on investment).[iii] However, it’s hard to put together a business case, and organizations would rather assume that there is a solid business reason for implementing AI than to take the time to determine benefits, feasibility, costs.
- Focus on implementing the technology without considering how the organization will get and use the AI results. As we all know and as Forbes states clearly, “AI should not be implemented for AI’s sake, but rather because it’s the most compelling solution to a specific business problem or opportunity.”[iv] It seems that today’s businesses often look for the shiniest new digital object and are afraid that they will be left behind if they don’t rush to implement it. In other words, they chase a solution, without giving much thought to the business problem solves. In addition, executives often undertake these AI initiatives without knowing what they’re going to do with the results they will receive from the machines.
- Mistakenly think that implementing AI projects will be easy. AI requires a difficult and usually painful organizational transformation. This digital transformation, as it is sometimes called, is a major cultural change that requires buy-in from all levels of the organization. It will be expensive. It means adding new positions like data scientists and eliminating others. Adopting Agile practices will seem easy compared to implementing an AI strategy.
- Don’t understand the questions being asked by data scientists—questions whose answers will be the basis for the predictive models. The AI insights come from machines that use algorithms that are based on rules. In order to develop these predictive models, data scientists ask the stakeholders for information, which is used to create the models. If the stakeholders do not understand the questions being asked or if they do not have a good handle on the insights they are looking for, the models will not meet their business needs.
- Seek to implement AI with an antiquated infrastructure and burdensome processes that do not support these initiatives. This is a major pitfall, and one reason why so many software vendors tout things like AI-ready software or an AI-ready infrastructure. Organizations are beginning to realize how important it is to build an appropriate infrastructure when implementing an AI strategy. This infrastructure needs to include not only the technical and data aspects, but the business processes that will need to support the new technology.
What can we BAs do to help organizations avoid these common pitfalls? There are many ways we can help organizations transform to the digital world and take advantage of digital technologies. In all things AI, we need to be trusted advisors. As strategic BAs, we can help the organization develop its AI strategy by recommending an approach that works for that specific organization. We can help implement those strategies, and we can recommend tools and processes that help ensure the AI effort meets expectations. Specifically, we BAs need to:
- Develop a business case for AI initiatives, which has to begin by stating the business need for this initiative, stressing the business problem or opportunity. We BAs are great at this—it’s part of our DNA. In addition, we need to include industry-specific benefits, keeping in mind that they might not all be financial. For example, in health care include such things as saving patients’ lives and more accurate medical diagnoses. In the financial sector include security, and risk avoidance. In retail emphasize providing a unique and dazzling customer experience. We also need to look at all aspects of feasibility, including the organization’s culture and readiness for change. And let’s not forget about tying AI to innovation, since it’s hard to be innovative these days without some element of AI.
- Help the organization determine what AI results they want to avoid the pitfall of having an ill-conceived solution without thought given to the results the organization wants, nor the decisions they’ll be used for. This is a key to successful AI projects. And we can play the translator role, helping business stakeholders understand the ramifications of the decisions they are making that will become part of the predictive models created by data scientists.
- Help ensure that the historical data is accurate. AI insights come from amassing huge amounts of data. If the data is not accurate or if the desired information has not yet been collected, the results will likely be worthless. BAs are good at this. Understanding data and the ramifications of bad data is one of our bailiwicks. We can inform the organization of the risks and ramifications of not spending the time and money to cleanse the historical data before it is amassed for AI applications.
- Reviewing the current state and recommend changes. We need to review the current business processes, as well as its infrastructure, existing and needed data, types of new roles and positions needed, and more. Again, understanding the gap between what exists and what is needed is something we BAs are good at. It’s a core competency.
- Finally and importantly, BAs need to recommend how to get everyone on board with this transformation. BAs need to help the stakeholders understand what happens if key stakeholders are not bought into this major change. And that transformational change takes time—sometimes lots of time.
So when BAs hear that AI is being implemented because it’s the cool, latest technology with the promise of huge profits, they need to say “hold on—not so fast. Implementing AI for the sake of AI—for sure not!”
[i] https://hbr.org/2019/03/the-ai-roles-some-companies-forget-to-fill, Harvard Business Review March 2019
[ii] https://www.gartner.com/smarterwithgartner/how-to-build-a-business-case-for-artificial-intelligence/, April, 2018, contributor Christy Pettey
[iii] https://www.forbes.com/sites/insights-delltechnologies/2019/04/02/how-to-build-a-strong-business-case-for-ai/#147960e67f1a, Forbes Insights Team, April, 2019
About the Author:
Elizabeth Larson, PMP, CBAP, CSM is Consultant for Watermark Learning and PM Academy and has over 30 years of experience in project management and business analysis. Elizabeth’s speaking history includes repeat keynotes and presentations for national and international conferences on five continents.
Elizabeth has co-authored five books and chapters published in four additional books, as well as articles that appear regularly in BA Times, Project Times, and Modern Analyst. Elizabeth was a lead author on the BABOK® Guide 2.0, the PMBOK® Guide 4th and 5th editions, as well as the BA Practice Guide. She was also an expert reviewer on both the BABOK® Guide 3.0 and the PMI standard in Business Analysis.
Elizabeth enjoys traveling, hiking, reading, theater, and spending time with her 6 grandsons.
After delivering the closing keynote of the 1st European Business Analysis Day in 2018 (see archive), Elizabeth will be speaking on the topic of Artificial Intelligence in the 3rd BA-DAY in May 2020. Here you can find her session.