What is Practical AI?
Practical AI (Artificial Intelligence) is the valuable application of intelligence rendered by machines. Practiced AI is rooted in present-day use cases, and is separate from the futuristic promises and predictions of what artificial intelligence may be able to accomplish.
The term artificial intelligence, coined in 1955 by Dartmouth math professor John McCarthy, has undergone changes in its scope since its initial use. As the field of artificial intelligence has expanded, two phenomena have played particularly important roles in our collective understanding of what it means:
1. The emergence of the AI Effect. This describes how as machines have become more intelligent, small accomplishments in the field drop off the radar of what is considered AI. The AI Effect has led to artificial intelligence as a term lending itself well to futuristic leanings, including those of:
2. AI in popular culture. Frequently the backdrop of post-apocalyptic thrillers, artificial intelligence in popular culture has painted prominent pictures that aren’t grounded in the reality of the day, including the fear-based rise of technological singularity, where AI will lead to runaway technological growth that overtakes and even displaces humanity; and tales of what AI may allow us to do in the future.
Practical AI, then, grew as a result of how the AI Effect and AI’s portrayal in popular culture led to a muddled, distorted purchasing and selling environment where vendors of AI-powered solutions felt the need to lean into and leverage pop culture’s use of the term, and customers found it increasingly difficult to separate the signal from the noise.
On the vendor front, this has led to many solutions providers AI-washing their product. On the consumer front, this has led to a sense of distrust and confusion because, as Sol Rashidi, Chief Data and Cognitive Officer at Royal Caribbean International put it, everybody wants to put their technological solution “under the AI umbrella just because there’s a bit more glamor to it.”
So then what is artificial intelligence?
As the image above shows, there’s no shortage of questions being asked about artificial intelligence. Unfortunately, the hyperlaxity of the term exists even in the field itself.
At Forrester’s CXSF 2017 event this year, for example, a collective gasped emerged from the audience when Pinterest’s CTO Vanja Josifovsk stated that artificial intelligence and machine learning are essentially the same thing. “I’ve been doing this for decades,” he told the audience. “And I see them as basically the same… just at different points on a continuum.”
The somewhat-agreed-upon definition of AI is that it is intelligence displayed by machines.
But when terms such as machine learning and deep learning are thrown around and often used interchangeably with artificial intelligence, it can become difficult to gain a fundamental understanding of one or the other.
Many writers of articles on each of the subjects assume their readers already have an in-depth knowledge of AI or are content with their relatively abstract understanding. And as mentioned with AI-washing, most vendors aren’t explaining the topics very well either.
The concept of Practical AI serves a critical role in separating the relevant from the simply interesting.
Let’s briefly explore the relationship between these terms.
How does artificial intelligence differ from machine learning and deep learning?
As you’ve likely gathered by now, heated debates exist about where these terms intersect or separate.
The image above represents the traditional view. Though the beliefs around this view seem to be changing slightly, it’s still a foundational model used by many universities and by speakers on artificial intelligence.
One of the more significant changes is as David Thieras points out in his piece at Forbes titled Machine Learning: The Evolution From An Artificial Intelligence Subset To Its Own Domain:
“It took me a while to wrap my head around it, given my earlier AI biases, but I’ve concluded that machine learning is now its own discipline, intersecting with both AI and BI in a very overlapped Venn Diagram.”
Why is Practical AI important?
While the definition of Practical AI will remain the same, its scope will shift and evolve in similar ways to the terms we’ve covered above.
Its importance rests on its immediacy and on its ability to provide the public (including consumers and vendors) a greater ability to determine the signal from the noise.
In an article at Gigaom, Rudina Seseri, founder at Glasswing Ventures and Entrepreneur-In-Residence at Harvard Business School, echoed words similar to Sol Rashidi:
“AI has now become a buzzword. Startups work AI into their pitches even if their businesses aren’t really oriented around the technology.”
Perhaps the best way to move beyond buzzword is to see a few Practical AI examples:
Examples of Practical AI
The most practical real-world applications of artificial intelligence, according to Erik Brynjolfsson and Andrew McAfee of MIT, fit into two categories: perception and cognition.
In regards to perception, voice commerce and image recognition are prime examples.
While Alexa, Siri, and Google Assistant have paved the way for voice commerce, its democratization is underway. Many online retailers, for example, are powering their on-site search functionality with technologies (such as Natural Language Processing) that can easily respond to longer, more expressive search queries. Beyond simply being a cool feature, there’s value in the practicality of it.
A recent study at Stanford showed that speech recognition is currently about three times faster than typing on a cell phone. This is sure to improve, and it means customers can more quickly find what they’re looking for (which can lead to big wins for the retail company).
AI, in continuously learning all facets of human speech, including which words matter in a particular query, is at the heart of voice commerce.
Similarly, image recognition is changing entire industries.
It’s helping doctors improve the accuracy and speed at which they can detect cancer, and it’s opening up new educational possibilities for everybody, everywhere—such as being able to snap a picture of a plant and instantly gain information about it.
And as with voice commerce, digital merchandisers are opening up search functionality by allowing consumers to search by photo.
In regards to cognition, AI examples tend to grab headlines and are often talked about for years to come, such as IBM Watson defeating Jeopardy champions and Google’s AlphaGo defeating the Go master.
But there’s also the quieter, more Practical AI manifestations: technologies allowing insurance companies to assess credit risk and more quickly process claims; and customer engagement platforms that can predict with stunning accuracy what you’ll want to purchase next, and display that item in real-time.
Both examples improve operational efficiency for all parties involved and can lead to time-savings for the customer and increased productivity for the provider.
Final thoughts on Practical AI
Ultimately, Practical AI is about what’s valuable right now. While peering into the future is exciting and can be valuable, it can also serve to distract us from seeing the use cases that are already available.
And when the seemingly infinite definitions of AI are all used with equal weight, it can appear as though a relatively simple chatbot CMS plugin has the power to take over the world.
In this sense, Practical AI can help establish an important baseline that improves communications in the consumer-vendor relationship.