Artificial Intelligence, Robotic Process Automation (RPA), Machine Learning (ML) and Deep Neural Networks (DNN) have been hyped significantly in recent years as transformative technologies. No doubt they will continue to have a major impact on many aspects of our lives in the years to come. Yet today, truly transformational application of these technologies is more the exception than the rule.
For example, in 2013 the MD Anderson Cancer Center launched a “moon shot” project to leverage IBM Watson to diagnose and recommend treatment plans for cancer. In 2017 after over $60 million spent, the project was halted because none of the plans had been used with patients.
Today, this technology, the data available for processing – and how to utilize it – has advanced. Many leading cancer centers now leverage AI in recommending treatment plans for patients based on multiple factors. The technology just wasn’t ready several years ago.
While the initial project was on hold, the cancer center’s IT group experimented with less ambitious applications of AI and RPA. These included making hotel and restaurant reservations for patients’ families, determining which patients might need help paying bills, and helping with staff IT problems. These more modest projects produced better results. The new applications resulted increased patient satisfaction, improved financial performance and a reduction of time spent on routine tasks.
Cognitive technologies are increasingly being used to solve business problems, but many of the most ambitious problems encounter setbacks or fail. What seems to work best is an incremental approach rather than starting with a transformative approach – with a focus on augmenting rather than replacing existing processes.
The main uses for these technologies include:
- Process automation. RPA today is more advanced than process automation tools of the past because the “robots” (code) acts more like humans. The new tools can consume and process information from multiple sources. Typical uses of this technology include: automatically transferring data from call center and email systems to update systems of record like customer information, or replacing lost credit cards, etc. This is currently the most common application of these technologies.
- Cognitive insight. This includes the application of technologies to detect patterns in large volumes of data and interpret their meaning. Essentially this represents the next evolution of the analytics capabilities that have been delivered with enterprise applications for years. We have already seen applications of cognitive insight in our daily lives: predictions of what customers will buy – personal ads, fraud detections, etc.
- Cognitive engagement. The most advanced application of these technologies. These include the combination of multiple aspects: natural language chatbots, intelligent agents, and machine learning for data analysis. Use cases include product and service recommendation with increased personalization, health treatment recommendations to provide customized care plans, etc.
Like many things, learning how to use these new technologies is itself a process. To get the most out of AI, RPA, and ML/DNN, organizations need to understand which technologies are most appropriate for the specific tasks at hand. As the technologies evolve and the uses for them are better understood appropriate applications will emerge. Opportunities for effective use of these technologies exist where an organization learns, and every place where there’s a feedback loop.
As with any business systems, start with the objective – the problem to be solved and the new capabilities to be delivered. Resist the temptation to use the technology as the solution and then attempt to find a problem.
McKinsey: April 2018
HBR: February 2018