Although vendor-written, this contributed piece does not promote a product or service and has been edited and approved by Network World editors.
The philosophical war between artificial intelligence (AI) and intelligence augmentation (IA) has been waged for more than half a century, with the focus shifting between the two as each has made important strides.
On one side, the AI camp believes the future of computing is autonomous systems that can be taught to imitate/replace human cognitive functions. A recent example is Google’s autonomous car, where the machine completely replaces human intervention and interaction.
On the other side, the IA folks believe that information technology can supplement and support human thinking, analysis, and planning, but leave the human at the center of human-computer interaction (HCI). Consider a car collision avoidance system that can help a driver prevent an accident, but doesn’t actually remove the driver from the picture.
The last two decades have witnessed AI’s rising fortunes, with the success of IBM’s Deep Bluecomputer, which beat chess grandmaster Gary Kasparov in 1997, IBM’s Watson defeat of Jeopardy champions Ken Jennings and Brad Rutter in 2011, and most recently, Google’s AlphaGo defeate of Go champion Lee Sedol this year.
These successes demonstrated the superiority of computers over humans in accomplishing a certain kind of undertaking. And following each, countless predictions emerge of the ascension of machines and the demise of the human. Vivek Wadhwa, a distinguished fellow at Stanford and Duke universities and business technology specialist, recently said, "In a decade or two you'll find that robots and artificial intelligence can do almost every job that human beings do. We are headed into a jobless future.”
AI, bots and the cloud
According to AI theorists, it’s not just game-playing computer programs that are poised to wrest control of your life. In the last few months Microsoft, Google and Facebook have all announced bot frameworks – software designed to automate tasks, like setting up an appointment or performing an Internet search.
Modern bots employ AI technology to process conversations (or text sessions), effectively replacing the human operators who typically stand behind these processes. Recent examples of bots and chatbots include Domino’s bot for ordering pizzas, Taco Bell’s bot for ordering food via Slack, and (the slightly creepy) X.ai bot which automatically schedules meetings with colleagues while posing as a human.
Experts predict bots will soon replace apps as the primary way we complete tasks. The simplicity of the bot promises to replace the rigid structure of the app; it’s easier and more intuitive to use. Instead of navigating through an app, you will simply speak to (or text) a bot and tell it what you want.
AI is finding its way into the cloud, as well, where Microsoft and Google are already purporting to change the way we work. Using the artificial intelligence behind Microsoft’s Office Graph, the Office 365 app Delve presents users with recommendations for documents and conversations they may want to view. Google’s recommendations in apps like Google Now use the AI embedded in the Google Knowledge Graph to present information it thinks users will want to see, including nearby restaurants, shops, and museums.
This is just the beginning. Expect to see an endless stream of AI apps aiming to solve every problem you have ever had.
So that's It for IA, right?
After 50 years, has IA finally been vanquished? Are we ready to relinquish control to autonomous cars, software bots, and AI-based recommendation engines?
The answer is yes … and no.
While AI will clearly play a larger role in our daily lives, it is not a panacea. AI-based solutions work best in structured environments where all relevant information can be considered and where the goals of the system are clearly defined – ordering a pizza, setting a meeting, playing chess.
In all these cases, while the number of possible outcomes that has to be considered may be enormous, the outcome can be predicted with a high degree of confidence (and can be tweaked based on user response to improve results in the future). This is exactly the situation where a powerful computer has an advantage over the human mind.
On the other hand, artificial intelligence is not well suited to situations where goals and inputs are not well defined; it’s here where intelligence augmentation will continue to play a major role.
Let’s look at an example. A salesperson focused on closing business relies upon many different systems to do their job. Email is the main source of information today, but others include SharePoint, Box, or Dropbox for documents; Skype, Slack, Yammer, Chatter, text messaging, and the phone for real-time communications; and business apps for order processing, trouble ticketing, and customer relationship management.
On any given day, integrating that disparate information is an exhausting task. AI-based machine learning systems can extract topics from messages on each of these systems and even match them across systems. But then what does the salesperson do with that information?
Here is where AI reaches its limits and IA excels: assisting the human operator in evaluating what action should to be taken next.
Should they contact the prospect to offer a discount, reach out to the internal support team for help in solving a customer problem, or research a competitor’s offering to develop a competitive comparison?
What the information worker needs is an IA-driven dashboard, app, or widget that aggregates information from different systems, extracts topics from each update, and them matches them across systems to create a common vocabulary for work activities. It should then assist the user to arrange the information in ways that make it easy to digest – for example, with intelligent filters and visualization tools. Thus, IA allows the worker to play with the information, to arrange and process it in a manner that best lets them decide what do to next.
This case of the salesperson is not an isolated one. There are many such business processes where the human will remain in the driver’s seat for years to come. The job of system designers is to continue to provide them with the best tools to deal with information overload caused by disconnected information coming from many disparate systems. And IA will continue to play a leading role in this challenge.
A virtual détente
Today’s information worker is inundated by inputs from an increasing number of data sources, and as a result they are turning to a growing number of cloud services to get business done: for storing and sharing documents; for completing transactions via CRM, HR, and business-specific apps; and for communicating with peers via social tools like Skype and Slack.
And since individuals and departments can sign up for whatever services they need (often without IT’s approval), the corpus of tools being used by an organization is becoming progressively diverse. Trying to make sense of this cacophony of signals is creating an information overload for workers, who now run the risk of dropping the ball instead of focusing on, and completing, important work endeavors. Simply put, it’s becoming more difficult to see the information forest for the data trees.
AI-based systems can help here as well. They can help make sense of this Data-Tower of Babel by creating contexts out of information emanating from different systems. But that’s as far as artificial intelligence can go, because what comes next requires human intervention – and that's where intelligence augmentation still picks up the slack.
The AI vs. IA war isn’t a war after all. They both have an important role to play in our future.
A co-founder and vice president for marketing and strategy at harmon.ie, Lavenda is a veteran of the high-tech industry, having co-founded Business Layers and held executive positions at V-Secure and WorkLight.