by James Taylor, CEO, Decision Management Solutions
Artificial intelligence (AI) has the potential to transform businesses, and adoption is on the rise. A 2018 McKinsey global survey shows that 47% of organizations say they have “embedded at least one [AI capability] into their standard business processes, while another 30% report piloting the use of AI.”
At the same time, AI failure rates are strikingly high, partly due to lack of a well-defined strategy or aiming too high initially and expecting dramatic outcomes. IDC estimates a 50% failure rate and Gartner’s estimate is higher than 85%.
In light of these developments, the team at O’Reilly, a world leader in technology and business training and knowledge, recently reached out to Decision Management Solutions to share our thoughts about how organizations can extract more business value from AI. As experts in the field of decision management and decision modeling, we focus on helping organizations in various industries leverage technology tools to drive better business decisions.
When it comes to AI, the problem is not that the AI approaches and tools used fail to produce an algorithm. The problem is that they fail to produce business value. Let’s take an example in the medical field. In a research setting, an AI finds more cancerous tumors and sometimes detects them earlier than medical staff reviewing the same scans. From a research standpoint, this is a successful AI project because the AI indeed found cancer and found it earlier. But until this AI is used to change treatment for patients, until it is integrated into a diagnostic and treatment protocol and shown to generate better or cheaper outcomes, it’s not successful from a “business” perspective.
This leaves many executives scratching their heads as to how to productively utilize AI in a way that effects positive change throughout their organizations. Currently, many organizations are implementing AI, but only in pilot studies or for specific, one-off projects. One of the questions that is top of mind is: How does this maturing technology provide true value?
Of course, every business is unique—there’s no one-size-fits-all way of implementing AI. If you’re a big, traditional company with a stable business, you have to do AI differently than the glitzy technology companies like Google or Tesla that can turn on a dime. Traditional companies—such as those in the insurance, healthcare, and financial services sectors—need to take an entirely different approach and stay true to their strengths rather than follow the lead of technology giants that are more comfortable with AI and have the resources for ambitious large-scale implementations.
We have found that the best approach is to start with a specific business problem and decide what a “better” business outcome would be. Then work backwards to determine what technologies and approaches are needed in order to get there rather than making AI implementation a purely technology-driven undertaking. In short, to successfully deploy AI and get beyond a pilot, companies need to approach it backwards, beginning not with their data but with the decision-making they want to change.
Our highly practical approach is known as Digital Decisioning, which involves creating a Decision Model that balances human elements with AI in the decision-making process. A Decision Model clearly spells out the role of AI in a business context.
Make sure everyone is on the same page
Start with a cross-functional team that bring together business, operations, technology, and AI professionals to ensure everyone has the same end goal in mind—better decision-making for a well-defined business problem. Use decision modeling to lay out the decision-making approach in detail. This will help everyone understand the nature and intent of the project in non-technical terms.
Use AI with other relevant technologies to drive better decisions
The focus is on business value, so AI can be used along with other technologies, which together contribute to improved decision-making. AI can be used in combination with data science and predictive analytics to help improve accuracy of a decision, explicit rules-based processing for regulatory and policy-based aspects of the decision, and human decision-makers to define more intangible, real-world interactions that are part of the decision-making process.
Focus on continuous improvement
AI works best when it’s used on a continual basis. In fact, this is at the core of how it works. You can use a Decision Model to see what data to capture about a given decision, integrate this decision outcome data with business results, and use AI model feedback to see how to keep improving results and spot potential problems early. It’s also important to bring in subject matter experts to review this data so they can make modifications and changes as they start to gain an understanding of what AI needs in order to do its work successfully. Pursuing that purpose in small incremental changes can add up fast. It’s the upward slope of improvement over time that you should be aiming for, not an immediate upheaval. Success is achieved when you methodically rely on an iterative process of continual refinements rather than a moonshot approach that expects swift and radical exponential change.
Interested in learning more about how AI can help fine-tune your business decision-making and provide measurable value? Check out:
• “How to Succeed with AI” by James Taylor
• “AI Orchestration Methodologies: Creating Fully Intelligent Decisioning Process Automation” by Pam Baker