AI

office working app computer

It seems everyone is talking about Artificial Intelligence (AI). It shows that AI is becoming more and more of a business conversation and that’s generally a good thing.

The main issue with AI is the spiel and frank misunderstanding in business today. There is also the usual bandwagon jumping due to the many analyst firms saying that businesses need to adopt AI. This is no different to the plethora of buzzwords that adorn the pages of media, journalistic and social, be it Blockchain, Big Data, or whatever.

There are absolutely use cases for AI adoption in business, though that use case should always be defined and understood first. However, in many cases businesses are adopting AI without actually understanding what it is they are adopting. This is usually due to slick salespeople promising the earth, or simply a buzzword chasing organisation.

When looking at AI it is vital to understand some of the key concepts and differences. For example non-symbolic AI and symbolic AI.

The significant majority of AI offerings in the market today sit squarely in the non-symbolic camp. This, in the most part, is what was called Machine Learning and is now badged as AI. From an AI perspective this is round the bottom rungs of the ladder to true artificial intelligence. It is primarily based on algorithms and statistical models, which require data upon more data from which to reach a conclusion. From a scientific perspective this conclusion would be deemed a tentative hypothesis. The reasoning being that the conclusion of such an approach is not explainable, or human readable, nor proven or disproven. It is a viewpoint based on the data to hand, which of course requires accuracy of both data and algorithms.

We have already seen major examples where such an AI approach has been implemented and subsequently removed as it did not live up to the expectations, nor performed with the accuracy of human knowledge. This last aspect is absolutely key!

With symbolic AI the approach is different. Symbolic AI seeks to replicate the approach a human would take in understanding the facts to hand, combined with residual knowledge from which to formulate an outcome. It does this through a combination which includes Knowledge Representation and Reasoning (KR&R), the reasoning aspect of which allows for inferences to be made about that knowledge and subsequent information. The KR&R is underpinned by ontologies, which define domain specific information , classes, attributes, relations, axioms etc. This fundamental approach means that a symbolic AI engine is able to replicate the approach a human would undertake in problem solving or decision making, but equally be able to show how any conclusion was reached. Thus giving greater potential for accuracy and also understanding and confidence therein. A symbolic AI system effectively starts with a hypothesis and through knowledge understanding, fact interpretation, inferences and confidence in such inference seeks to prove or disprove the hypothesis, from which an action can be undertaken.

Where we think of AI we need to think beyond the acronym, into an understanding of the differences between approaches. For example consider that non-symbolic AI is entirely focused on telling us about the data and patterns in the data; whereas symbolic AI Is entirely focused on integrating human knowledge (individual, group, organisation, industry) together so it can perform reasoning over that integrated knowledge to solve complex problems.

The preceding explanatory section is important for understanding why we think organisations who do not have this level of knowledge should look to people like Byte™. Even if looking at other consultancies, or talking directly to vendors, hopefully this provides a little in your armoury to hold sensible conversations and identify the wheat from the chaff.

At Byte™ we have a thorough understanding of this area, from which we can aid any organisation on their pathway to AI adoption, ensuring that decisions in this space offer true business value and achievable outcomes. We want to see business outcomes and the potential from AI adoption realised. This will only happen if that organisation has a thorough understanding of AI at its disposal. We hope you will see Byte™ as that.