Looking at recent headlines, it is easy to see that AI technologies are close to going mainstream. Actually, one might say that AI is already mainstream, and everyone is using AI technologies one way or another. The biggest problem that AI faces, of course, is that of the shifting goal post. No one really knows what AI is, so essentially, AI has become this distant dream that is inspired by great fictional creations such as Daneel Olivaw , Skynet, and the like.
So the goal of what would be considered AI has shifted continuously over the years. Not too long ago, Optical Character Recognition (OCR) was considered a bold new frontier for AI. And of couse, once we cracked OCR, everyone realized that it was all about some smart algorithms and pattern recognition, and AI really had nothing to do with it. We’ll really need AI to beat humans at Chess. Wait, that’s done already. But Go is more challenging. We’ll really and truly need AI for a computer to ever beat a human at Go. And now that’s done too. A computer beat the top Go player.
What has perhaps changed of late is the visibility of AI. If AI was once limited to deployments at supercomputers for highly specialized functions, it is now becoming commonplace. And the front end—the interface with which we access AI—has become highly accessible as well as humanized. It’s certainly been a long journey from the psychotherapist Eliza, developed in the mid sixties, to today’s Siri and Alexa. And these new interfaces, with remarkable Natural Language Processing (NLP) prowess and huge troves of data to dig into are getting noticeably smarter and more capable with each iteration.
Perhaps it was the increasing visibility and pervasive presence of AI that prompted tech visionaries such as Elon Musk and Bill Gates to issue a call against the use of AI in warfare. The ethics of using AI in warfare aside, the fact that this group thought AI was advanced enough for such a warning to come is itself telling.
In precise scientific terms, there is no clearly defined concept for AI. In common usage, anything that is currently beyond our current computing capabilities is often labeled as AI. For all that, AI, of course, takes different forms. Some significant examples of AI available widely today include bots on Reddit and other platforms that summarize news articles and the like for the convenience of others, the chat bots on Slack, and of course, the new crop of intelligent personal assistants such as Siri, Alexa, Cortana, and Google Now. Behind the scenes, AI is powered by a range of different technologies including NLP, big data and pattern recognition, neural networks focused architectures, and more.
And as AI matures, it is moving beyond programmatic computing to cognitive computing and deep learning. With these new capabilities, it is not surprising that AI is expected to have a bigger role to play in the Enterprise.
While it is still early days for AI in the Enterprise in some ways, it is easy to see that there are a number of domains in which AI can play a significant role. To be sure, AI is already at work in a many domains such as healthcare (patient diagnosis and illness prediction), finance (stock trading, financial analyses, etc.) logistics (optimization of resources in samples operations such as shipyards, warehousing and large scale retail enterprises, etc.) ERP, and more. But these, one might say, have been specialized, focused deployments of AI. Some more focused areas for AI include:
Enterprises that deal with large troves of data can benefit from intelligent automation of their report generation activities. This type of report generation goes far beyond running a little script that will perform a fixed set of actions and deliver output in a pre-configured format. Instead, these AI algorithms will mine data and recognize noteworthy patterns, learn to identity and account for changes in data based on patterns such as seasonal changes, possible connections between two distinct areas of the business, and so on.
Quality control, especially in the field of customer service, is a significant preoccupation for most large businesses. Currently, quality control for customer service relies on tools such as feedback requests to customers and statistics such as average call time, resolution rate, and so on. These methods require significant manual effort while still offering only a limited range of insights, that too with a time delay. The deployment of AI can help in this scenario, as the system can then process large amounts of data in real time to identify the kinks in the customer service flow.
Human resources typically deal with large troves of data, yet most of it continues to be unstructured or unorganized. Deploying AI capabilities in HR will make optimum use of such data, to accurately identify problem areas related to attrition, recruitment delays, employee satisfaction or growth paths, and so on.
The folks at Narrative Science have put together some insightful research that is worth citing in this context. I’m listing some key points from their research here, though the entire infographic is certainly worth a look.
- 80% of Enterprise executives believe that AI increases productivity and crates jobs.
- Over 48% of the companies currently use AI for automated communications
- Just about 6% of the companies use AI for automation of repetitive tasks.
- Almost 32% of the respondents in their survey highlighted the importance of voice recognition and response powering AI.
This last point in the Narrative Science report is particularly interesting, in the light of the great strides made by Amazon’s Alexa and Apple’s Siri. This space—AI powered personal virtual assistants—is highly exciting, and Google too is joining the fray with its newly announced Google Home. And as these AI assistants get refined, smarter, and more efficient with their widespread adoption in the consumer space, it is inevitable that these tools will also find their way into the Enterprise world. I can already imagine an enterprise version of Alexa that helps with scheduling and conferencing; coordinates on and creates travel plans; attends meetings, takes minutes, and emails them to all participants; or simply sits at the office entrance, so that employees can enter office saying ‘Hello Alexa’ instead of swiping their ID cards or fingerprints, and Alexa opens the office door for each one of them using its voice recognition capabilities.