“Artificial Intelligence” is a major buzzword phrase right now. It seems like every new product or service offering is pushing the fact that they leverage AI, machine learning or neural networks to make their product better and cheaper. It’s certainly reasonable to believe that automation will improve quality and lower cost; it’s been happening for decades in physical manufacturing. However, it’s just over the past few years that we’ve seen this trend beginning to dominate the virtual world. Marketing strategies, customer interactions, artistic productions and much more are all being automated by leveraging AI.

Unlike physical automation where we can clearly see a robot performing tasks that we understand, virtual automation is more difficult to comprehend. We don’t really know how the human mind works and so we tend to treat virtual minds with the same mystery. I find that, for many people, when we’re speaking about virtual automation we can replace the phrase “Artificial Intelligence” with “Magic” and communicate the same level of understanding.

Imagine the following phrase:

“We’re going to optimize your customer’s journey by leveraging artificial intelligence to produce dynamic messaging”

Sounds good, right? If we make one small change does the sentence still tell you the same information?

“We’re going to optimize your customer’s journey by leveraging magic to produce dynamic messaging”

For many people that I speak with these two sentences are basically the same. While it isn’t practical for everyone to understand all the intricacies of AI, it is helpful to understand some of the basics so you can make educated decisions. This also can keep you safe from getting caught up in the hype and believing the impossible.

Learn the Lingo

Let’s start by defining a couple of the key terms in this field. Firstly, there is “Artificial Intelligence”, this is a broad term that refers to computers doing things that are like what humans do, such as learning or making decisions. AI, by itself, tells us very little about what a thing is. It’s like using the word “vehicle.” This could be a car, boat, unicycle or rocket. Each has its own purpose and excels in different applications. However, chances are that none of them are the supercomputer Skynet intelligence many people imagine.

In fact, in most cases Artificial Intelligence is not actually all that intelligent, it must be taught. Most AI leverage some form of Machine Learning to do this. ML is a specific technique for implementing AI. The basic principle behind ML is to define a measurement of success and rules for behaviors, then let a computer program iterate over these rules, trying semi-random behaviors until it finds what works best for the given rules. This all may seem a bit vague, but the key takeaway is that Machine Learning is a form of rapid evolution. The AI tries and tries again, discarding the methods that fail and modifying the ones that succeed. This might make sense to see in action, here is a great example of an AI being created to play Flappy Bird.

This works great for applications where it is possible to try repeatedly. For example, Machine Learning was a critical aspect of Alphabet’s AlphaGo program. It could play against itself millions of times until it improved beyond the ability of humans. This method is much less effective in applications that can only be attempted a small number of times. Without iteration the AI doesn’t have enough time to do any learning.

Dozens of other terms exist that refer to specific methods of implementing Machine Learning, these details are more technical than this introduction will touch on. For now, when you read phrases like Neural Networks, Bayesian Learning, Deep Learning, etc. it is safe to substitute the term Machine Learning in most practical applications.

Machine Learning is one way to build AIs, but there are non-learning methods as well. These implementations have no capacity to grow on their own and are completely dependent on a human giving specific instructions. This is quite limited since it is time consuming to build and can require expensive ongoing enhancement and maintenance to meet changing needs. 

There are also many speculative future implementations where an AI possesses a theory of mind. For now, these exist only in science fiction.

What is AI doing right now?

Today, AI is used in many applications. It is improving search results, making spam filters smarter, driving cars, trading stocks, and keeping people healthy. The list of things AI can be used for is extremely impressive, but what about for the average business?

In the world of marketing we can see many of the factors that make AI and ML viable. A/B testing is a common practice and provides a basic set of rules to try. Success of a strategy is measured easily with metrics we already collecting. It shouldn’t be a surprise that major platforms like DoubleClick are already offering tools to help marketers leverage AI for themselves.

For content creators, AI can be harnessed to augment your content creation schedule. Major publishers like the Washington Post already supplement human reporters with virtual ones. While that content is quite utilitarian, AI is not limited there, it can be used to create music or model.

On the eCommerce front, brands like Disney are using AI to enhance discoverability and improve the customer experience by tagging, categorizing and guiding their user’s journey across many different brands, products and experiences. They can use ML to discover what works and what doesn’t, often finding patterns that a human would miss.

Personal interactions are also very possible. Chatbots for customer service can be trained to supplement human staff – often dealing with customer inquiries in a more dynamic and efficient fashion than a rigid set of menu options found in most automated systems, while also reducing the load on customer service staff.

The sheer number of options can be both amazing and intimidating. Like any new service or product, it’s easy to get overwhelmed or caught up in the hype. 

If you think you have a use for AI in your business or if you’d just like to know more, please reach out to us, we’d love to help!

Jon Bailer

About Jon Bailer

Jon Bailer is the Director of Technology at Liquid Interactive. Jon oversees the activities of the Technology department, bringing over a decade of experience implementing custom technology solutions for a wide variety of clients.