At one point, artificial intelligence (AI) was confined to science fiction movies and our imaginations, but now it seems to be everywhere. What exactly is this technology and how does it work? Should it be part of your app?
In this article, we explore artificial intelligence and help you decide whether it’s worth your investment.
What is AI?
Artificial intelligence is the practice of making computers think like humans. It’s when a machine processes data and develops new conclusions that weren’t intended by the machine’s programmers.
Famous cryptanalyst, mathematician, and computer scientist Alan Turing created a test (appropriately called the Turing Test) to determine if a machine can demonstrate human intelligence. In the test, a computer tries to impersonate a human while talking to other humans. If the real people fail to identify the computer in more than half of all attempts, the machine passes the test and is declared intelligent.
However, the Turing Test isn’t perfect. The results aren’t perfectly clear. For instance, a computer might fool humans when asked specialized questions, but fail when asked open-ended questions. You don’t need original thought to trick a person, just really good conditional logic.
And that’s the crux of the debate over artificial intelligence: the term is almost always misused, especially by marketing teams. When they say “AI,” they are almost always referring to algorithms.
An algorithm is a set of step-by-step instructions, written by humans, for computers to follow. We see algorithms in real life all the time: cooking recipes, instruction manuals, every “how to” post on the web, etc. Computers are really good at following algorithms, even if the instructions are complex.
Following an algorithm, however, is not creativity. It does not seem like a human thought because the computer isn’t coming up with new ways to solve a problem. It is bound by its instructions.
The misuse of “AI” is complicated by machine learning, another technology with another slightly misleading name. Machine learning is the practice of letting machines draw from vast sources of data to compute through their algorithms. But this also isn’t intelligence, even if the machines store their collected data for future processing.
The Truth About AI
What does this mean? Basically, true AI doesn’t exist yet. Despite their ability to process mind-bogglingly sophisticated algorithms, machines can’t generate original thought.
“Today’s tools for businesses involve mathematics, statistics, machine learning, deep learning and big data—with better machines than in the past,” writes Max Simkoff in Scientific American. “But what is so often referred to as AI doesn’t actually involve an artificial form of intelligence. […] Understanding this is crucial for businesses that want to take full advantage of the opportunities new technologies have to offer and build defenses against future competition.”
“AI,” therefore, is just a marketing word that refers to automated algorithms. Manufacturing robots, self-driving cars, robo-advisors, chatbots: These are all just rule-based systems that achieve complex goals, but they don’t think. The humans who wrote the complex conditional statements are the only intelligences.
“Will we ever see AI?” you might be wondering. Possibly, but there’s no good answer at the moment. But right now, the gap between machines and human thought is pretty wide.
Let’s put aside the fact that what we’re calling AI isn’t true AI. For the sake of argument, we’ll use the marketing definition of AI: sophisticated algorithms plus machine learning.
The applications for AI are remarkably impressive. Today, there are countless AI-based services for businesses of all sizes. These tools help you collect data, gather insight, and add lots of value for your customers. This explains why companies in every sector are investing in AI. According to the 2019 CIO survey by Gartner, the number of companies implementing AI technologies has grown by 270% in the previous four years.
Here are examples of how AI is powering new applications:
Learning user behavior patterns
Natural language technology
Personalization of content
Predicting user behavior
Ad relevance (and conversion rate)
With those tools, we’re getting countless apps that solve complex problems with ease. Here are several companies that are pioneering with AI:
Conversion.ai - An AI that writes high-converting copy for ads, emails, websites, blogs, listings, and more.
MeetElise - An AI leasing assistant who immediately responds to leads, schedules tours, and follows up to maximize closings.
Terzo - An AI-powered PDF importer that scans contract PDFs and automatically adds the contract data to the vendor management platform.
Frase - An AI content brief tool that helps you prepare the perfect content for your topic.
Wysa - An AI-powered conversational agent that helps improve mental health.
AI in Your App
All of this begs the question: Does AI belong in your applications?
Before you start slapping AI tech into your systems, it’s important to ask yourself if it adds value. Don’t include it because it’s trendy. Don’t include it just because you want to call your app “AI powered.” It’s never smart to use technologies because you feel like you’re supposed to.
As always, start with your strategy. What is your application supposed to accomplish? What problems does it seek to solve? Can any of those problems be solved in a better way?
Then consider the best way to solve those problems. Is AI a reasonable solution? If so, it may be worth the investment, but only so far as it addresses the users’ core issues. For instance, if only 3% of users operate the search function in your app, you probably don’t need an AI-powered search feature. It just wouldn’t move the needle enough to be worth your time and cost.
Further, when it comes time to actually add artificial intelligence to your application, you don’t have to start from scratch. We already have existing tools, platforms, and frameworks that can add AI to your app (with some integration and customization work, of course). For example:
Microsoft Cognitive Toolkit (CNTK) is a training algorithm for machines to learn like humans. It’s used to create machine learning models. It’s highly customizable because it lets you select your own networks, algorithms, and metrics.
Amazon ML is a machine learning framework with tools to create intelligent and sophisticated applications without any code. It works with desktop and mobile apps, and can connect applications to the cloud.
PyTorch is a library and a scientific computing framework that’s known for its efficiency, speed, and numerous pretrained modules. A lot of big companies use this one, like IBM, Yandex, and Facebook.
But if AI doesn’t support your strategy (your app’s purpose), then adding it would be pointless. In fact, it would be a distraction that prevents you from spending time and resources on more important updates and improvements.
We haven’t achieved true artificial intelligence yet, but we might in the future. What we call “AI” right now is actually just algorithms that solve complex problems, but not genuine intelligence.
Still, the tools we’re calling “AI” can be quite valuable for all sorts of applications. They can solve complex problems, manipulate data, and even draw conclusions. But AI is only valuable if it serves your app’s purpose. If it doesn’t help your users solve their problems, it’s not worth your investment.