There are widespread misconceptions about Artificial Intelligence (AI) and these extend to its powers and its abilities. Potential users may have unrealistic expectations of what they can expect to see it do.

For some, AI conjures up images of robots, while others expect it to be able to solve all problems and master numerous manual processes and tasks without training or configuration. The truth is that although AI has advanced exponentially over the last decade, it is still really in its infancy.

The gap between reality and marketing is exacerbated by the existence of some technological products that claim to be artificially intelligent when they aren’t. This is the world of ‘AI-washing,’ which is becoming something of a phenomenon in the business world.

Some so-called 'AI solutions' are merely data aggregators. At best, they are offering machine-learning software products, whereas at worst they’re just providing software algorithms. This has been occurring for a while; in the summer of 2017 the research and advisory company Gartner stated: “many technology vendors are now ‘AI washing’ by applying the AI label a little too indiscriminately.”

Why is it happening? Sad to say, it seems mainly to be due to commercial opportunism. I know people who believe that using the label of AI will - even falsely - elevate the value of some software, no matter how dated it may be. It makes it appear to be cutting-edge innovative technology, thus warranting a higher price and generating more revenue per product sale. Vendors who have not kept up with technical innovations or who have not invested in true AI may think that they can hide behind a plethora of misconceptions about AI by branding their products as AI-powered. This means that companies may be purporting to use AI but are in fact using Robotic Process Automation or Big Data analysis.

Caveat emptor - let the buyer beware! What’s on the tin may not be what’s inside it. If in doubt whether you’re looking at some genuine AI, here are some questions to ask the software vendor.

• Can your software be continually taught or trained on multiple 'use cases' by a business user without any AI knowledge or data analytical skills? [A 'use case' is a list of actions or event steps typically defining the interactions between a role (known in the Unified Modelling Language as an actor) and a system to achieve a goal. The actor can be a human or other external system. Even though FinTech and RegTech use the same technology (AI, blockchains and so forth), their 'use cases' are very different from one another.]
• Does the AI software continue to learn from what it is being taught?

In both cases the answer you are looking for is yes.

You could also ask the vendor the following.

• How do you define AI?
• What AI technology does your solution use?

Big Data and the growth of the Internet of Things have created a fertile breeding ground for AI "application solutions" that work in more than one sector, ranging from financial services to health and education. It is crucial to the development of company strategies to understand the effect of AI on a business, as it affects their market(s), their customers’ needs and the changing nature of competition in business. Used correctly, AI can transform a business process or issue for the better.

There is no doubt that genuine AI has revolutionised technological operations by being able to create instantaneous and automated human-like intelligence that functions very well. They help to automate business processes, pull unstructured data from documents, make companies more efficient by saving time and costs, reduce risks, provide advanced real-time analytical insights that help people make decisions and function like a human brain. Genuine AI is intuitive, powerful and accurate.

In terms of regulatory compliance in financial services – for example automating the KYC processes to counteract money laundering and terrorist finance – 'Utopia’ is an AI system that harnesses machine learning and natural language algorithms. The AI engine should not be static but trainable to understand any regulation regardless of its country of origin. Also, once the system has learnt a regulation, it should be simple – using a straightforward 'text input' interface – to teach the engine to understand any differences or changes in regulation. It might take a person weeks to understand and be trained to appreciate changing regulations, whereas AI can do it in a matter of hours.

Since it can take months to complete compliance assurance processes manually, the business case for embracing AI compliance software's automated efficiencies, cost savings and analysis is undeniable. Such platforms are available and are infinitely superior to the illusory masquerade of the AI-washing brigade.