Artificial Intelligence will shape the future of business more powerfully than any other technology innovation this century. Any business not understanding AI will soon find themselves left behind, disrupted by competitors using systems to make decisions more quickly. As with almost any major technology disruption there are many factors that complicate the successful deployment including new, immature technology, lack of experienced practitioners, and known system patterns. This is causing many business to struggle with how to get started with using Artificial Intelligence. There is no shortage of research and pilot projects underway but most business are not getting the expected results.
Working with many Dell Technology customers across many industries including manufacturing, healthcare, and finance I have identified three major components of successful AI strategies:
- Identify and prioritize those existing functions that could be done better by a computer system
- Identify the data you will need for the computer systems that will perform those functions
- Create an AI technology taxonomy including the infrastructure and software frameworks you will need
Identifying the current tasks that could be done better by a computer system using AI requires close collaboration between business leaders and IT. Understanding most computer systems using AI today are typically most successful with a single task. For example, gathering product for an order, analyzing an x-ray, processing a loan application can be completed faster, and more cheaply by a computer system using AI today. Other tasks can be greatly improved by augmenting employees with AI computer systems. For example, I have a customer that is improving sales by prioritizing opportunities for their sales force based on their customer social media activity.
Once the opportunities and expected outcomes are prioritized, the next step is to identify the data needed for the computer system to apply artificial intelligence. If you do not have the right data your intelligence will be suspect and less reliable. Many times the business has the data but it must be conditioned differently for AI systems. If the data is not being collected, a plan must be developed and executed to collect or acquire the data. Often this is when my customers realize they need a chief data officer role to manage and govern data inventory. As data becomes more important to operations, storing, protecting, and securing it becomes more important to not only IT but the business.
Many IT teams will immediately jump to developing a technology taxonomy as the first step. While it is important for IT to experiment with new technology, the AI technology taxonomy needs to be influenced by the use case. There are several popular AI platforms including TensorFlow, Premonition, and Rainbird as well as a number AI as a service providers such as IBM Watson, Microsoft Azure Machine Learning, and Google Cloud Prediction API to name a few. Understanding your use case, expected outcomes, and data types will allow you to narrow the AI platforms and services you use. These services utilize different models, and algorithms to optimized for different use cases.
In future posts I will share the successful AI use cases, data architectures, and the technology taxonomies that my customers have deployed. The businesses that have followed this strategy have been much more successful in realizing the tangible benefits of AI including higher production efficiency, differentiated services, and user experiences. Businesses that don’t start with these three steps are rarely successful, take much longer to see any business benefit, and spend more money to achieve those results.
Wow wasn't expecting such an informative article. Just had to take the time to thank you, for this was what I was searching for. You just gained a new follower today. Thanks and happy holidays!
Posted by: Victoria the Computer Technician | 12/27/2017 at 06:47 PM