Socrates’ view of human nature is summed up in the doctrine: to be knowledgeable is to be virtuous. Researchers in artificial intelligence have relied on philosophy (thinking about thinking), in addition to examining smart animals and engineering intelligence in machines. Tech companies like Apple and Google are already trying to build programs capable of reproducing human capabilities to learn and solve problems.
They’re building, as Socrates would say, virtuous machines. A great example is the iPhone 7. The iPhone’s camera has long been a differentiator, but in the most recent iteration Apple has introduced an image signal processor (ISP) enhanced by machine learning.
Bokeh, or the art of blurring out certain elements in the background of images, can be easily done with a standard DSLR camera. But on a smartphone, with software that sees photos as a collage of numeric values for each pixel, this is a much more difficult effect to execute. Machine learning is a must—the way Apple and other technology companies are ‘teaching’ programs to pick out elements in a photo by feeding them thousands upon thousands of examples.
Ultimately, just like humans, AI will need to draw on a constantly-growing database of information. An intelligent program should read historical data, analyze it for patterns, and be able to classify what it sees. Without a database to learn from and then call upon this information to match with new data, a program cannot really “learn”.
For most enterprises, practical use of AI is not yet feasible. The actual solutions on the market are not very accessible, by and large. A good example of AI for the masses is Google introducing machine learning to the G Suite, formerly known as Google for Work. By shaving seconds off delays at every level, Google is trailblazing user-friendly AI. Not everything need be as complex as IBM Watson!
It’s clear, however, that AI is ascendant. It’s one of the frontiers in technology being pushed aggressively, as we saw from Microsoft at Ignite and SFDC at Dreamforce. What will determine who leads the AI efforts in the next decade and beyond? It will be the companies best equipped to “teach” using massive amounts of data.
If you doubt the ever-increasing value of data, look no further than Microsoft’s recent acquisition of LinkedIn. Make no mistake: this was the purchase of knowledge, more than just a Rolodex. Since its inception, LinkedIn has amassed a treasure trove of information on who the world’s professionals are, who they interact with most, their career paths, and their topics of interest.
There are fascinating synergies between this information set and the productivity tools Microsoft offers, and Microsoft has moved to take advantage. The greatest tech in the world will not be intelligent—or, perhaps, virtuous—unless it is “taught” using robust and relevant data sets.
My use of the word ‘teach’ is deliberate, because a student learns in much the same way as an AI program would. First, by reading (acquiring knowledge passively), then internalizing (analyzing the information actively), and then applying the learnings to solve new problems.
In this case, the AI is the student, learning how to attack new problems. The insights and information collected as analytics comprises the book. Much like in real life, the more books you read the more knowledgeable you become. A student who can read but always reads the same two books will simply be exposed to fewer ideas than a student who can read and can access the Library of Congress.
If we extend the metaphor a bit, we can see that the companies with the highest potential to master AI will be the ones with the biggest databases and the richest analytics. The AI future belongs to the ones with comprehensive datasets that draw on historical data to help programs master topics from all angles. With this in mind, first-mover advantage is a real and very powerful differentiator.
Google, for instance, was the first company to truly understand the value of crawling the Internet to collect information. Because of that core foundation—that rich library—their search is now the dominant service. With all this historical data on people’s searches, webpages and associated web links, Google hopes to predict the best sites for you from the first keystroke.
SalesForce built the best database of customers-sales interaction in the industry, and they are now increasing its value by cataloging, analyzing it and extracting insights to improve the sales process for both salespeople and customers. Indeed, thanks to that early commitment to analytics, they are in position to create AI to better match sales representatives with prospects, predict customers’ questions to help providing the right information at the right time, drive the sales discussion in the most efficient way, etc.
There are more examples across industries, and AI will play a bigger role in data-driven business in the coming years. How will first movers like Google, SFDC and others continue to solidify their advantages? Can smaller players catch up build virtuous machines to close this gap?
(About the author: Isabelle Guis is chief strategy officer at Egnyte)