Artificial intelligence is still often associated with Terminator-like beings and sci-fi dystopias – but AI is currently a hot topic and this trend, luckily, comes with a lot of educational value. It’s getting easier to find reliable information about AI and learn how it works, yet people tend to be distrustful towards new technologies. It was much easier to get used to other innovations once we understood how they work but with AI, “How?” is still a big question. We don’t know how it reached its conclusion, so how can we trust it? There are some approaches addressing this issue, such as LIME (local interpretable model-agnostic explanations), which I mentioned in my article about the things you must consider before implementing AI in your business.

According to research by O’Reilly, difficulties in identifying appropriate business use cases remain one of the major hindrances to AI adoption. Before you implement artificial intelligence in your business, it’s good to know the possibilities and limitations that there are. Today, I want to show you what AI does and how that translates into practical business use cases.

How does AI work?

Before we move on to the possibilities that artificial intelligence offers, we should have an understanding of what it is. AI is a concept of machines being able to perform tasks typical of human intelligence, including recognizing objects and sounds, understanding language, planning, and problem-solving. This is a very broad definition and since we often assume that AI uses its “intelligence” just like humans do, many people believe that artificial intelligence is a human-like machine. While building machines that resemble humans is definitely a field that science is exploring, what we deal with now is specialized AI or “weak AI”.

So AI can be divided into 2 types of main classification:

Weak AI

Weak AI, or narrow AI, is the one we’re working with today. It’s focused on one specialized task. In this case, the system is limited to the tasks it’s programmed to perform. These are still sophisticated uses of technology: autonomous vehicles, image recognition, recommendation engines or intelligent assistants (e.g. Siri) but they are only able to handle a limited set of pre-defined tasks.

Strong AI

Strong AI, or general AI, is a term that refers to machines that are able to perform any task just like a human being. There are currently no real-life examples of strong AI but many organizations are working on the development of this concept. Every instance of weak AI will contribute to the building of strong AI.

Then, AI is further divided based on its functionalities.

Types of AI

Perhaps the biggest buzzword among these types is machine learning. ML means all the techniques and processes that bring AI closer to understanding human cognition. You can often hear people use the names “artificial intelligence” and “machine learning” interchangeably, but in fact, they are not the same. Machine learning is a subset of AI, a way to “achieve” artificial intelligence, which allows the systems to learn. You can read more about AI and ML here.

When we move further down this line, you can also see deep learning. This, in turn, is a subset of machine learning and is a method that utilizes artificial neural networks to allow systems to train themselves, much like humans do. There are many types of neural network architectures, and they include (the more and more popular) convolutional neural networks (CNN) that can be used e.g. in diagnosing diseases from medical scans, recurrent neural networks (RNN) that can be used e.g. for assessing the risk of fraudulent credit card transactions, and generative adversarial network (GAN), which received some media coverage thanks to creating realistic pictures of people that don’t exist.

What AI can do for business

There is a whole variety of business use cases for artificial intelligence. AI can predict call volume in call centers to support staffing decisions, detect fraudulent credit card transactions, forecast product demand, classify customers, filter spam email, predict customer behavior, recommend products that the customers will enjoy, but also detect faulty products on a production line, generate captions for images, power chatbots, provide language translation, and even help diagnose patients.

It may seem like AI systems are extremely complicated and difficult to achieve. While solutions providing patient diagnosis are in fact very complex, many organizations utilize simpler AI-powered tools that bring great results. Predictive analytics, NLP, recommendation engines, and computer vision are some of the most popular technologies. How do they help businesses? Let’s have a look at some examples.

Predictive analytics

Predictive analytics provides an assessment of future trends. Using the available historical data, it:

  • reduces churn by identifying the clients who want to leave,
  • increases sales by suggesting what the customer wants to buy and determines how much they will buy,
  • identifies employees who are likely to leave the company,
  • predicts demand for resources, product, and inventory,
  • identifies the risk of breakdowns, failures, malfunctions, and errors.

A real-life example: the system we built for a Polish telecom company.

Our client was struggling with growing customer churn, which is a common pain among commodity businesses. The client was unable to prevent churn due to a number of factors, including an ineffective customer retention strategy. The aim of the first stage of our project was to reduce churn by 2 percentage points in the customer segment where churn was highest. The pilot was scoped for 10 months and included a 360 view of each customer, predictive models with churn predictions and with product recommendations, as well as tools delivering suggestions on how to take care of churn-prone customers. The project was finished in less than six months with the result of churn dropping by 20%.

Recommendation engines

Recommendation engines can be utilized in a variety of areas, including movies, music, books or other products. These engines will:

  • personalize offers based on the customer’s earlier choices and reviews,
  • suggest other videos, songs, books, etc. that the customer will like,
  • increase conversion rate,
  • improve customer experience,
  • increase retention and loyalty.

A real-life example: recommendations by Netflix.

The recommendation system developed by Netflix is probably one of the best recommendation systems in general. 80% of the content watched by their users is based on the algorithmic recommendations. The system analyzes vast volumes of data to determine what other films or series you will like. The system decides the probability of a user liking a given film based on a few factors, including the user’s interactions with the service (such as viewing history and ratings), other members with similar tastes and preferences, information about the titles (genre, categories, actors, release year, etc.). It also analyzes data about the user, such as the time of day they watch, the devices they use to watch Netflix, and how long they watch. This way, the algorithms provide users with a list of personalized recommendations, along with the probability of the user liking the suggested content.

Natural language processing

NLP lets companies better use language data thanks to the effective processing of large amounts of voice or text messages. Natural language processing:

  • extracts specific data from long texts (e.g. articles, books, bills),
  • automatically processes invoices, orders, contracts,
  • identifies the writer’s emotions and recognizes hate speech,
  • improves customer experience with chatbots and virtual assistants.

A real-life example: better writing with Grammarly.

Grammarly helps fix grammatical mistakes and avoid overused words. The company developed a system that combines rules, patterns, and artificial intelligence techniques like machine learning, deep learning, and natural language processing to improve their users’ writing. Grammarly’s system was trained with the use of a huge collection of texts that were organized and labeled by humans. To provide accurate suggestions, the system keeps learning from feedback from users. E.g. if a lot of users hit “ignore” on a given suggestion, adjustments are made to make the suggestion more accurate.

Computer vision

Computer vision (CV) allows machines to see, identify, and process images to provide the right output. CV can be used to:

  • verify people’s identity,
  • allow users to virtually try on clothes,
  • equip automotive vehicles with information about surroundings,
  • diagnose conditions and illnesses,
  • monitor factories or other sites,
  • scan crops to determine their condition.

A real-life example: checkout-free shopping with Amazon Go.

Amazon Go is a new shopping experience offered in a few US Amazon stores. It’s a checkout-free solution where all you do is grab the products of your choice and go. You are then charged for your shopping through your Amazon account. Amazon uses computer vision, sensor fusion, and deep learning for this solution. The “Just Walk Out” technology they developed automatically detects when products are moved from the shelf or returned to their place, and it keeps track of the products in a virtual cart.

What industry can benefit from AI?

Any industry. AI will learn anything that you want it to, so there’s a lot of flexibility when it comes to its use cases. What’s important to remember, however, is that you don’t have to follow the tech giants. You don’t have to use AI the same way that Amazon, Facebook or Netflix do. If you want to implement AI in your business, you should follow the money, and not somebody’s lead. See how your competitors use AI, think about the processes that you need to improve, match your business needs and goals with technology.

If you don’t know how AI could be used in the area of your interest, here’s a cheat sheet:

Though every organization can benefit from AI, there are some industries that are already being transformed thanks to artificial intelligence. The areas that are currently experiencing the biggest changes include healthcare, finance, transportation, education, cybersecurity, as well as e-commerce and customer service.

The cheat sheet you can see above includes only some of the use cases of artificial intelligence in given fields. I haven’t listed all the possibilities – there are just too many. Not only that, AI is a field that’s constantly developing, so we’ll see new use cases in the near future.

Right now, the healthcare industry uses AI to collect individual patient data, among other uses. This not only allows to keep their history up-to-date online (can you imagine a world where every doctor you go to knows about your conditions, allergies, etc?), supports diagnosing, but also provides tools for better management and monitoring of chronic diseases such as diabetes, asthma or heart conditions. Medical practitioners utilize AI-powered solutions to diagnose conditions and perform surgeries. And we, as patients, can keep track of our data thanks to IoT. We can even consult virtual medical assistants! And remember, we’re just at the beginning of the road.

Finance is a sector that was eager to implement AI in its early days. AI solutions are now used to manage investments, collect financial data or anticipate changes in the stock market. Remember the times when the New York Stock Exchange was a place full of brokers shouting at the top of their lungs? That’s the picture we saw in every movie that had Wall Street in the background. Times have changed. There aren’t nearly as many people around. Trading is now computerized, so their job has changed as well. Banks already test solutions that utilize image recognition to analyze and verify documents.  The customers of these institutions are provided a better customer service thanks to AI-driven chatbots.

When it comes to education, AI helps make it more accessible and individualized. Right now, the level and methods of teaching depend not only on the country you live in but also on each educational institution. With AI solutions, teachers would be able to provide a more personalized approach, without dedicating additional time to the task. The collaboration of teachers and AI may bring some fantastic benefits in the future. Adaptability and individual approach, online classrooms available to students that would otherwise miss. What’s more, AI could provide tools for students with hearing impairment with real-time subtitles, and those students who speak different languages with real-time translation. Solutions reading facial expressions to provide feedback for teachers are already available but not yet popular in the education sector. In the future, however, solutions like that will not only help teachers see how efficient their methods are but also which students are struggling – a thing that’s so easy to miss in an overcrowded classroom.

Again: these are only some examples of AI’s business use cases. We could spend hours talking about solutions for chosen industries but in the end, it all boils down to one thing: your business. It’s about your needs and your goals. AI can be implemented in any part of the company to optimize processes, support staff in their everyday work, and, ultimately, help your business grow. It’s not about what the others do, it’s about how you can use AI to achieve your goals.