Like “big data” and “social media” before it, the term “artificial intelligence” has become so buzzworthy at this point, it’s largely lost meaning.
If everything seems to be powered by A.I., that’s because many companies are desperate to be perceived as leaders in machine learning (or deep learning, or natural language generation, all of which fall under the AI. umbrella)—even when they’re not.
It’s understandable. Artificial intelligence is an increasingly powerful force in the world, even as our grasp of what AI is and does continuously evolves. I like how the software engineer Liza Daly recently put it: “Artificial intelligence tends to mean whatever it is that computers can’t quite master yet.”
So it makes sense Jeff Bezos, the Amazon founder and CEO, spent a good chunk of his latest letter to Amazon shareholders focused on artificial intelligence. The letter is worth reading in full. It contains of all kinds of pleasing Bezosisms—“disagree and commit” deserves its own article, really—but I want to focus on the section Bezos devotes to artificial intelligence, which he describes as a big trend but one also “strangely hard for large organizations to embrace.”
“At Amazon, we’ve been engaged in the practical application of machine learning for many years now,” Bezos writes. “Some of this work is highly visible: our autonomous Prime Air delivery drones; the Amazon Go convenience store that uses machine vision to eliminate checkout lines; and Alexa, our cloud-based AI assistant.”
Here’s where it starts to get more interesting:
“But much of what we do with machine learning happens beneath the surface," he writes. "Machine learning drives our algorithms for demand forecasting, product search ranking, product and deals recommendations, merchandising placements, fraud detection, translations and much more.”
And then, when Bezos gets into what all this means for Amazon Web Services, Amazon’s cloud services platform, is where it gets really interesting:
“Inside AWS, we’re excited to lower the costs and barriers to machine learning and AI so organizations of all sizes can take advantage of these advanced techniques,” Bezos writes.
He goes on to describe how Amazon’s cloud-services clients can use the company’s pre-packaged deep-learning frameworks—including the systems that power the Amazon Echo; Amazon Polly, the company’s text-to-speech program; and Amazon Rekognition, its facial recognition software.
Clients have access to these technologies through a simple API, Bezos says, meaning developers for a range of companies can tap into Amazon’s suite of AI programs without having any machine learning expertise themselves.
This is a big deal for a few reasons. Mainly, because it means Amazon is enabling countless organizations to track its users more precisely than ever.
AWS is huge. It is, for starters, the backbone of the commercial web. (That’s why, when AWS has a server problem, it seems like the entire internet is coming apart at the seams.) It reported a stunning $12.2 billion in sales last year, and more than $3 billion in profit. Giving all AWS clients easy access to advanced data-tracking tools means Amazon is making the baseline for corporate surveillance online much more sophisticated.
This may be exciting news for businesses that want to follow and analyze their customers and potential customers more closely. But for people concerned about individual privacy, this is not so great. (Amazon did not respond to my request for an interview.)
Of course, many of the biggest companies that use AWS already run in their own high-level data-tracking operations. (Remember, AWS clients include McDonald’s, Netflix, Airbnb, Adobe, Capital One, GE and Pinterest, to name a few.) It was in Bezos’s shareholder letter last year he boasted of AWS’ stunning client base: “More than a million customers from organizations of every size across nearly every industry,” he wrote.
AWS helps power a massive swath of the global economy—across markets and industries all deeply vested in collecting and sharing detailed data about individuals and their behaviors. Yet, there’s clear incentive for companies to leverage machine-learning technology beyond tracking individual behaviors. There are all sorts of applications for a computer that can be trained to recognize patterns.
Bezos included this lucid explanation in his letter: “Over the past decades, computers have broadly automated tasks that programmers could describe with clear rules and algorithms. Modern machine-learning techniques now allow us to do the same for tasks where describing the precise rules is much harder.”
All this is a reminder that the public-facing conceptions of how businesses use artificial intelligence—the device sitting in your kitchen that responds to your voice, or the drone dropping an Amazon package on your doorstep—are only a teensy slice of what a company means when it says it’s using AI.
Instead, much of the corporate world’s algorithmic fine-tuning can be boiled down to learning exactly who customers are and how they behave. That’s already happening in the background of every business transaction you make—especially online—with remarkable precision.
“Though less visible, much of the impact of machine learning will be of this type,” Bezos says in his letter, “quietly but meaningfully improving core operations.”
“Watch this space,” he adds. “Much more to come.”
Much more to come, certainly. But we’re not likely to learn very much more by simply watching, when so much of what artificial intelligence does is invisible from the outside.