AI with two conversation bubbles and a question mark printed with a fingerprint
All stories

What do people mean when they say AI?

This might surprise you, but the most important thing about AI isn’t its use case. The most important thing about AI is how predictive it is. What does it mean for AI to be predictive, and why does it matter?

AI lets you feel like you’re in Vegas, but all the time

If you’re smart about using AI for your business, you know it’s not a magic pill to solve all your maladies. It’s like you’ve suddenly developed the ability to see through cards at a high-stakes poker table in Vegas. At its core, AI helps you use data to “gamble better,” or make more informed decisions. Having AI that helps you make the right bet is so important.

When people talk about business applications for AI, they’re usually talking about prediction. If AI is predictive, it’s interpreting data correctly, making the right bet, and giving you the right guidance. The more predictive AI is, the more often it gives you the right guidance. The more often it gives you the right guidance, the more value it provides.

There is no AI solution that will be right 100% of the time — but different types of AI are smarter than others because they learn from their mistakes. There are many solutions on the market calling themselves AI that don’t learn or get more predictive over time. So how can you tell if the AI you’re investing in is smart enough to actually help you level up?

AI ≠ Smart AI

In general, I find it useful to think of AI as a spectrum from least to most advanced. The most advanced AI is also the most predictive, and thus the most likely to show value. Where AI falls on the spectrum is defined by how it learns from and acts on data.

Basic AI = Process Automation

At the least advanced end of the spectrum are AI solutions that automate processes. This article from the New York Times gives great examples of the “least intelligent form of AI” in B2B technology today, “following simple rules and making yes-or-no decisions.” There’s usually a high degree of human involvement at this level; people help set the rules that dictate the yes-no decisions. When the list of rules updates, it’s because humans did it manually.

This is the area where non-AI platforms masquerading as AI tend to slip in. To be “official” AI, the rules need to be based on probability and statistics learned from an initial dataset. Do you know if the AI you use is based on statistics and probability? Or is it hard-coded by humans? More human involvement means less accuracy. This is great to understand if you want to know whether the product you’re thinking of buying is likely to work for you.

Smart AI = Learning Loops

screen shot of the Waze app

Source: Thanks, dude463!

Further along the spectrum we have machine learning, a subset of AI. Machine learning uses data to learn from its own decisions. These learning loops help make the product more predictive as time goes on. The more data and learning loops you use, the better it will work for you. Would you rather use printed-out paper directions? Or Waze, which uses traffic data to make better recommendations on directions in real-time?

Learning loops require less human intervention to achieve higher accuracy. Textio’s Director of Engineering is great at explaining the many ways that Textio does this. If you don’t have time to read it, just know that it’s much more advanced than a static list of rules.

Deep AI = Deep Neural Networks

At the far opposite end of the spectrum we have deep learning, again an even more complex subset of machine learning. At its most advanced deep learning (also known as deep neural networks) uses layers upon layers of algorithms to understand multiple features and react accordingly.



Textio customer NVIDIA has some great examples of how they’re helping customers solve some of the most complex problems in computing today. Consider all of the information that a self-driving car has to take into account to make the right decisions as it drives. Computer vision recognizes lines, shapes, depth, and more to classify images, detect objects, and inform the car to respond safely.

Why should you care?

It probably doesn’t surprise you that as a language company, Textio cares a lot about words and their meaning. But you should care too! In language, and especially tech language, definitions and meanings change over time. For Textio and our customers, getting clear about the value of augmented writing means getting clear on the definition of AI.

For our partners outside of tech, for example those in HR, what does AI mean to them? If you’re investing in or building any emerging technology, bridging this gap is incredibly important. To understand what results you’ll see, you’ll need to truly understand the technology first.

All stories
Get stories like these delivered right to your inbox.