Sergey Nivens - Fotolia
- Maribel Lopez, Lopez Research
There's tremendous fear that artificial intelligence will replace jobs, but this dystopian view doesn't acknowledge that AI software can help augment the business outcomes employees shoot for in many organizations.
The proliferation of mobile devices in the enterprise has been a major driver of AI growth because both technologies aim to improve the productivity of end users. Digital assistant technology, for example, has matured since assistants such as Apple's Siri first appeared. Plus, mobile, social and the internet of things have created a vast amount of new connected data sources. Cloud computing, big data storage and new processing technologies enable IT to collect and analyze this data better and at a cheaper cost than in the past.
AI is a set of disciplines that include machine learning, deep learning and cognitive computing. Machine learning uses recognition of patterns to allow software to predict outcomes without being explicitly programmed. It relies on algorithms that can receive data and use statistical analysis to predict an output value within an acceptable range. Deep learning uses artificial neural networks, or systems that rely on sending signals to gather information similar to the way brain neurons do, to mimic the operation of the human brain. Lastly, cognitive computing -- using deep learning -- applies knowledge from cognitive science to build systems that simulate human thought processes.
Three ways AI software helps improve business processes
Here's how IT can build intelligent, data-driven organizations using AI software:
Analyze data to predict outcomes. Machine learning can uncover new insights in big data analytics by finding patterns across multiple data sets and generating predictions using the patterns extracted from the input data. For example, a retailer could use machine learning to identify potential customers for a targeted marketing campaign and predict the type of campaign that will deliver the best outcome for a specific customer base. Organizations could use AI software to optimize product turnover by matching up weather data, historical purchases and current sentiment analysis to determine the best product inventory for a region.
Provide the right information, at the right time. Often, an employee can't solve a problem promptly because he or she is missing data or has incorrect information. Machine learning can predict what data that user might need, based on context such as where the employee is or what they're trying to do.
A simple example is Google Now, which reviews a person's calendar and traffic patterns to prompt the user to leave for his next appointment at the appropriate time. AI software could merge data from multiple systems of record and engagement to predict potential reasons for a customer support call, route the call to the best agent and provide options to reward the customer for their loyalty.
Assist employees with the next best action. Machine learning and deep learning can improve business outcomes by suggesting potential follow-up actions. For example, AI software could proactively alert a salesperson to a possible delay in a customer's shipment, outline proposed alternatives for the client and suggest a new product offering based on how the company uses its existing service. After reviewing thousands of images and medical research papers, AI can help a doctor define treatment for an illness by recommending the most effective treatments and new therapies that have the highest potential for success based on the patient.
Prepare for enterprise AI technology
With access to timely insights, employees can make better decisions, improving overall company performance. AI software can help human resources staff evaluate more résumés and identify top candidates, or it can help telecom providers improve network coverage and minimize outages by predicting potential points of failure and congestion.
Organizations must lay the foundation for enterprise AI technology now. First, IT should collect as much data as it can afford to store. Additionally, a company should create a specific data tagging structure that will help AI software be able to locate and analyze specific data at a later date. For example, an insurance company may store images of hurricane damage with meta tags such as hurricane, damage, Florida and roof. A consistent set of meaningful tags will make it easier for AI algorithms to learn the characteristics of an image and classify other untagged photos.
Still, most organizations lack data scientists and mathematicians, making it difficult to deploy AI tools. To minimize these challenges, vendors such as Amazon, Google, IBM and Microsoft offer cloud-based AI services to support natural language processing, image recognition and video analytics -- as well as a wide range of APIs that IT in any industry can use. Cloud services can be a good way to test the benefits and limitations of AI software.
It's also important to frequently reevaluate AI offerings, because the technology and vendor landscapes are changing rapidly. For the organizations that prepare for AI technology, it provides the opportunity to eliminate mundane repeatable tasks and focus on delivering innovations that matter to their customers and employees.
Dig Deeper on Enterprise mobile app strategy
Why transparency in AI matters for businesses
Machine learning and predictive analytics work better together
Understanding how deep learning black box training creates bias
6 basics of storage automation, predictive analytics and AI