AI Technologies

AI Driving the Next Innovation Cycle in Enterprise Software

Friday, September 21, 2018

Artificial intelligence (AI) is a term used to describe software applications that are able to mimic the cognitive capabilities of humans, such as the ability to learn and solve problems. Although the scope and formal definition of AI have been disputed over time, the concept is not new, having been founded as an academic discipline in 1956. Since this time, AI has experienced multiple cycles of extraordinary hype and disappointment as the associated technologies have been unable to deliver the promised real-life benefits. Yet with the confluence of a few recent trends—most notably, the explosion in global data volumes, declining cost and widespread availability of compute resources, and advances in computer hardware—we believe the pieces are now in place for AI to unleash a new wave of disruption.

As machines have become increasingly capable, tasks that were once considered AI have been removed from the definition (e.g., optical character recognition). The capabilities that technology analyst Bhavan Suri believes are most appropriately classified as core AI technologies today include machine learning, deep learning, natural language processing, and computer vision. “It is important to note that these technologies are not mutually exclusive” he said, “as most are powered in some way by machine learning.” We provide a brief description of each below.

  • Machine learning technologies leverage several techniques to extract knowledge or patterns from data, including regression analysis, cluster analysis, genetic algorithms, and neural networks. Machine learning essentially deals with the construction of algorithms that can learn and make predictions from data. In this way, they are able to overcome the need for explicitly programmed instructions by making data-driven predictions and decisions. This allows machine learning systems to become smarter as new data is ingested and consequently make better predictions.
  • Deep learning is a variation of machine learning. It expands standard machine learning by allowing intermediate representations to be discovered, which allow more complex problems to be tackled and others to be potentially solved with higher accuracy, fewer observations, and less cumbersome manual fine-tuning.
  • Natural language processing (NLP) uses algorithms to convert human language input into other representations. Although natural language, speech recognition, and text-to-speech have been possible for over a decade, recent advances in deep learning technology have significantly increased the accuracy and performance in these areas. Common application areas for these include virtual assistants, customer service, consumer products and games, hands-free control, healthcare, and military uses.
  • Computer vision technologies capture, process, and analyze digital images to derive information and recognize objects. In other words, these technologies essentially decode the meaning and context of images. Computer vision underpins the development of applications across many industries, including self-driving cars, autonomous drones, automated retail stock checks, and augmented/virtual reality.

Primary AI Application Areas

“The potential applications of AI are far-reaching,” Suri stated. “However, our research suggests that five primary application areas have emerged: 1) bots, 2) virtual assistants, 3) conversational AI platforms, 4) analytics and predictive analytic models, and 5) smart objects, sensors, and environments.”

  • Bots are software applications that automate various tasks by leveraging predefined rules or sophisticated algorithms, which often involve machine learning. For example, bots are often used to invoke other services or applications, such as ordering a pizza. The defining characteristic of bots is that they carry out the task for which they were built without human intervention.
  • Virtual assistants are essentially bots that can progressively handle simple natural language queries, act on simple natural language commands, and engage in limited dialogue with the user. Virtual assistants have emerged as one of the fastest-growing areas of AI and exist in many forms, including voice bots, text bots, and SMS bots. Examples of virtual assistants are Siri, Cortana, and Nina.
  • Conversational AI platforms are able to converse in natural languages with people via touch, gesture, speech, keyboards, and other means. Two forms of user interfaces that leverage conversational AI are voice assistants (e.g., Cortana, Siri, Google Now, and Echo), which parse voice commands, and chatbots (e.g., Facebook M and Slackbot), which respond to text. These platforms use NLP to parse the intent of human language.
  • Analytics and predictive analytic models are being applied by organizations in various departments, including IT, marketing, claims, customer service, and production. Advanced analytics solutions that leverage machine learning technology are enabling real-time automated event detection, decision support, risk modeling, and customer profiling, which drive revenue lift, loss avoidance, and improved customer experiences.
  • Smart objects, sensors, and environments encompass AI integrated with robotics, smart buildings, and internet of things (IoT) applications. AI technologies, such as deep neural networks, can master vision, sound, environmental, and other sensory input to add value in all of these areas.

Early Adopters Realizing Tangible Business Value
Although we have witnessed significant advancements in AI technologies over the past few years, commercial adoption remains in its infancy. Still, research suggests that AI is driving measureable business value for early adopters. “As part of our standard research process, we speak with hundreds of organizations and their employees on an annual basis,” Suri said. “and the feedback we have received over the past year has generated compelling evidence in support of this conclusion. Most notably, we have observed that early adopters tend to outperform their competition from both a growth and margin perspective, which appears most commonly due to their ability to deliver better customer experiences and leverage the technology to drive improved decision-making when it comes setting prices, identifying up-sell/cross-sell opportunities, optimizing logistical matters, and minimizing operational costs.” While Suri’s views are partly predicated on anecdotal survey work, more formal third-party evidence supporting his conclusions has started to emerge over the last year or two.

The best example to date is a 2017 study by McKinsey, which found that serious adopters of AI have current profit margins that are 3%-15% higher than the industry average in most sectors and significantly higher projected margins than other companies.

Analysts Jason Ader, Maggie Nolan, and Ralph Schackart also contributed to this report. For a copy of this report or for more information on companies from our technology coverage list, please contact your William Blair sales rep.

News Alerts

Stay connected to your favorite publications and news features.

Subscribe Now