The Use of Artificial Intelligence in Decision-Making
The Use of Artificial Intelligence in Decision-Making
Read this article in order to find out how CEOs should be implementing AI.
Join the DZone community and get the full member experience.Join For Free
The hype around the use of artificial intelligence in decision-making might make you think it could pilot your company automatically, talk to your suppliers, chase late invoices, and open parcels arriving in the mail room, all while making you a nice cup of tea.
In practice, AI is a precision tool that should be used judiciously to achieve specific goals. Implementing it takes foresight and a vision, along with a healthy understanding of the technical challenges involved.
Until CEOs understand the implications and requirements surrounding the use of artificial intelligence in decision-making, enterprises will not be ready to jump into machine learning in a systematic way. Here are three things that every CEO should understand before they tackle AI at the strategic level.
1. AI Must Be Aligned With Your Goals
In a McKinsey survey of 3,000 executives, 41 percent said that they are uncertain about the benefits of AI. AI can help your business by automating relatively simple tasks that would take humans 30 seconds or less to complete. It can also look more deeply at data to find patterns that humans may never catch. These are two very different capabilities, and CEOs must learn how to map them to their businesses by understanding their goals for AI.
Is your goal to reduce bottom-line expenses by automating simple tasks like sorting cucumbers on a production line or reading basic loan agreements? Or perhaps you are a hospital executive hoping to improve medical outcomes. You might monitor patient movements using wearable devices to predict the immediate risk of a fall. Or you might assist doctors in spotting and diagnosing potential health problems on MRI images.
Regardless of your goal, it will be important to use AI to support the user's experience. A doctor may benefit from diagnostic recommendations, but time constraints in a busy practice may only make this useful if delivered directly into a medical record during a patient consultation. A retail customer using natural language AI to help choose clothing will have a different set of requirements. The use of artificial intelligence in decision-making must fit workflows and formats that make sense for users. This brings user interface, business analysis, and workflow design into play.
2. Your Organization Must Be Data-Centric
AI thrives on data. The neural networks that are typical of AI systems learn to make better decisions using vast amounts of historical data. They then produce models that organizations must constantly refine as new data comes into the organization. Where will this data come from, and are you ready to provide it?
Most organizations have existing IT infrastructures built up by multiple teams over many years. The result is a fragmented information landscape. Data resides in different systems that don't talk easily to each other. Power structures can exacerbate this problem, creating political tensions that cause people to hold onto their data.
Breaking down these human and technological barriers takes a mixture of leadership and investment in technology. A CEO committed to strategic AI will lead from above, enlisting key allies in the organization to help unify its data architecture. A more fluid exchange of data throughout the organization can help with many projects other than AI. It's a foundational practice for the modern, digital-first organization.
3. You Should Pursue Strategic Partnerships
AI may eat and drink data, but it must be the right data. Understanding which information you need in order to train your machine-learning models requires multiple kinds of expertise. The first is domain knowledge. You need people with an intimate understanding of your organization's operations and how it uses different kinds of information to achieve specific outcomes. The second kind of expertise involves data science. Data scientists work with data engineers to extract, manipulate, and prepare data for AI workloads. If a data science platform isn't at work from this point, a developer will work to retrain, automate, and deploy models for consumption and productizing for consumption by a range of applications or systems.
This constellation of skills-from data science to software development to business process design will be a tall order for many companies as they struggle to transform their businesses. Combine this with the need to engage cloud services for compute-intensive neural network training, and many CEOs may find themselves overwhelmed.
Partnering with product and service providers that have a track record of navigating the AI design, development, and deployment process is a proven way to help overcome these AI hurdles and drive your business to success. Good partners will have the technical and business understanding to assist with the use of artificial intelligence in decision-making. They will also advise you on how to overcome barriers to data aggregation and management.
In such a nascent and fast-moving field, it pays to have specialist expertise to help guide you in your journey. The results could get your business well ahead of the curve, while others try to grapple with the problem in-house.Read more about the trends in artificial intelligence that could affect your enterprise.
Published at DZone with permission of Robert Hryniewicz , DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.