It’s easy to get excited about AI projects. Especially when you hear about all the amazing things people are doing with AI, from conversational and natural language processing (NLP) systems, to image recognition, autonomous systems, fantastic predictive analytics, and pattern and anomaly detection capabilities. However, when people get excited about AI projects, they tend to overlook some important red flags. It is these red flags that cause more than 80% of AI projects to fail.
One of the biggest reasons why an AI project fails is that companies don’t justify using AI from a return on investment (ROI) perspective. Simply put, it is not worth the time and expense given the cost, complexity, and difficulty of implementing AI systems.
Organizations are rushing to move beyond the exploration phase of AI adoption, jumping from mere “demo” to simple proof of concept to production without first assessing whether the solution will provide any positive return. One of the main reasons for this is that measuring the ROI of an AI project can be more difficult than initially anticipated. All too often, teams are pressured by senior management, colleagues, or outside teams to start their own AI efforts, and projects advance without a clear answer to the problem they’re actually trying to solve or the ROI to be seen. When companies struggle to develop a clear understanding of what to expect when it comes to ROI for AI, misalignment of expectations is almost always the result.
Missing and skewed ROI predictions
So, what happens when the ROI of an AI project doesn’t line up with management’s expectations? One of the most common reasons why AI projects fail is that the ROI is not justified by investing money, resources, and time. If you are going to spend your time, effort, human resources, and money implementing an AI system, you want to have a well-defined positive return.
Even worse than misalignment of ROI is the fact that many institutions do not initially measure or define ROI. ROI can be measured in many ways from financial return such as income generation or expense reduction, but it can also be measured as return on time, diversion or reallocation of critical resources, improved reliability and safety, reduced errors and improved quality control, or improved security and compliance. It’s easy to see how an AI project can provide a positive return on investment if you spend $100,000 on an AI project to get rid of $2 million in potential cost or liability, then it’s worth every dollar spent to reduce liability. But you will only see this ROI if you actually plan for it in advance and manage that ROI.
Management expert Peter Drucker once said, “You cannot manage what you do not measure.” The process of measuring and managing ROI for AI is what differentiates those who see positive value from AI from those who end up scrapping projects for years and millions of dollars in their efforts.
Boil the ocean and chomp on more than you can chew
Another big reason why companies don’t see the ROI they expect is that projects try to go too far at once. Repetition and agility best practices, especially those he uses Best practice AI methodologies like CPMAI We clearly advise project owners to “Think big. Start small. Repeat often.” Unfortunately, there are many unsuccessful AI applications that have taken the opposite approach by thinking big, starting big, and iterating infrequently. One example is Walmart’s investment in AI-powered robots for inventory management. In 2017, Walmart invested in robots to clear store shelves, and by 2022 they had pulled them from stores.
Apparently Walmart has enough resources and smart people. So you can’t blame their failure on bad people or bad technology. Instead, the main issue was a poor solution to the problem. Walmart realized that it was cheaper and easier to use human employees who were already working in stores to complete the same tasks that the robot was supposed to do. Another example of a project that does not return the expected results can be found through various applications of the Pepper robot in supermarkets, museums, and tourist areas. Better people or better technology won’t solve this problem. Rather just a better approach to managing and evaluating AI projects. Methodology, guys.
Adopting a step-by-step approach to running AI and machine learning projects
Have these companies fallen into the hype of technology? Were these companies just looking to have a robot roam the halls looking for a “wow” factor? Because being awesome doesn’t solve any real business problems nor does it solve a pain point. Don’t do AI for AI’s sake. If you only use AI for AI, don’t be surprised when you don’t have a positive ROI.
So, what can companies do to ensure a positive return on investment for their projects? First, stop doing AI projects for the sake of AI. Successful companies adopt an incremental approach to managing AI and machine learning projects. As mentioned earlier, methodology is often the missing secret sauce for successful AI projects. Organizations now see benefit in using methods such as Cognitive Project Management Methodology for Artificial Intelligence (CPMAI), Building on decades-old data-centric project approaches such as CRISP-DM and incorporating agile approaches to well-established best practices to provide fast, iterative sprints for projects.
These methods start with the business user and requirements in mind. The first step in CRISP-DM, CPMAI, and even Agile is figuring out if you should go ahead with an AI project. These methodologies suggest alternative approaches, such as automation, direct programming, or even just more people may be better suited to solving the problem at hand.
AI Go No Go analysis
If AI is the right answer, you need to make sure that you answer “yes” to a variety of different questions to assess whether you are ready to embark on your own AI project. The set of questions you need to ask to decide if you want to move forward with an AI project is called “AI Go No Go” analysis and this is part of the first phase of the CPMAI methodology. The “AI Go No Go” analysis causes users to ask a series of nine questions in three general categories. For an AI project to really progress, you need three things in alignment: business feasibility, data feasibility, and technology/implementation feasibility. The first of the three general categories asks about the feasibility of the business and asks if there is a clear definition of the problem, whether the organization is actually willing to invest in that change once it is established, and whether there is sufficient return on investment or impact.
These may seem very basic questions, but often these very simple questions are skipped. The second set of questions deals with data including data quality, data quantity, and data access considerations. The third set of questions is about implementation including whether you have the right team and skill sets required, the model can be implemented as required, and that the model can be used where it is planned.
The hardest part about asking these questions is being honest with the answers. It’s important to be really honest when addressing whether you want to move forward with the project, and if you answered “no” to one or more of these questions, it either means you’re not ready to move forward yet or you shouldn’t move forward at all. Don’t just go ahead and do it anyway because if you do, don’t be surprised when you waste a lot of time, energy and resources and don’t get the ROI you were hoping for.