CompTIA’s AI in Business – Top Considerations Before Implementing AI Guide helps business leaders and AI practitioners plan how AI can be used to improve profitability and operating efficiency. Despite artificial intelligence being a part of the technology landscape, many companies still struggle to integrate AI processes to achieve their business goals. AI’s transformative potential can only be realized when AI practitioners and business decision makers understand how it can improve business processes.
CompTIA’s AI Advisory Council has created AI in Business: Top considerations before implementing AI. This guide is meant to assist both business leaders and AI practitioners in deciding how to best use AI to improve their operating efficiency and profitability. It provides a quick overview of the major AI technologies and their implementation barriers. The guide then details 15 key questions that businesses should ask when embarking on an AI journey. This is based on the experience and expertise of council members.
Why AI and Why Now
Rama Akkiraju is an IBM Fellow and one of the contributing members. She says success stories from multiple industries are inspiring AI exploration by more companies.
Lloyd Danzig, chairman of the International Consortium for the Ethical Development of Artificial Intelligence, believes there is good reason to do so. He believes that more efficient algorithms, coupled with advances in computational power, open the door for solving problems that were previously unsolvable. He said that businesses can now use data to make better predictions, operate more efficiently and deliver a better customer experience.
This has been most evident in customer services operations, particularly with chatbots. Akkiraju said that IT operations systems management is another area that has seen positive traction. However, many applications in the healthcare industry have not scaled up because of the complexity and wide-ranging nature of areas such as radiology or oncology. Small startups have had more success with AI solving granular problems.
The First Steps and the Next Steps
Asking the right questions is key to any attempt at AI implementation. Danzig stated that the best next steps depend on the business type and their needs. After reading the guide and answering its questions, these steps should be clearer.
Danzig stated that every business and every use case is unique. “The Best Practices document can be used as a starting point to learn and gather context-specific information.
Akkiraju stated that some readers may be able turn to an internal data team to address the questions. Some companies, especially smaller ones, may choose to hire consultants to help with the assessment. However, everyone must address the one critical success factor that is not covered in the guide: expectations.
Expectation setting and expectation mismatch are two of the biggest mistakes in AI-infused solutions implementations. Overpromising can lead inevitably to disappointments. Akkiraju stated that AI vendors must be open and honest about what’s possible, what isn’t possible and how long it will take to realize value. “What’s possible in month one and five could be different.” A well-designed AI solution can learn and improve from user feedback. Companies might mistakenly believe that they will be able to create a system that automates every aspect of their business. AI systems often require time and iterations to learn from users and adapt to each company’s environment in order to achieve the required accuracy and deliver value.
This is true for AI vendors and practitioners, Akkiraju states. They panic when early results don’t meet expectations and then change their course.
“Builders of AI products must have a long-term perspective to solve more difficult problems with AI. It is important to adjust the course and refine the use cases based upon feedback. However, frequent product roadmap changes won’t solve difficult problems. To solve difficult problems, you need to invest in support and stay focused for long periods of time. She said.
Start small, build on that
Akkiraju stated that the key to an early win is a small beginning. An important insight can be gained from a time-boxed proof of concept. A small concept can make big strides if it is run by an external vendor or in-house data scientist team.
“Experiment with small use cases to better understand your strengths and areas for improvement. Create a roadmap that will outline how vendors or companies will address these areas and expand on the initial findings.