A simple framework to understand AI.

In brief:

• Artificial Intelligence (AI) is an umbrella term used to refer to several technologies - understanding their differences is important in order to optimise your technology investment.


• In this article, we provide a simple framework to understand AI software in terms of autonomy and intelligence.


• We define and examine four types technology that organisations may categorise as 'AI': Business Intelligence; Advanced Analytics; Robotic Process Automation; and Cognitive Computing.

The term 'AI' has become an umbrella for a variety of technologies that serve different purposes. Understanding the different flavours of AI is important in order to optimise your technology portfolio. Here is a simple framework to demystify AI, which you can use to evaluate the investments in AI technology needed to support a particular strategy:

A framework to understand AI.

Think of AI applications along two dimensions:

• Solutions that give people information to make better decisions (Inform) versus those that make decisions by themselves (Automate).


• Technologies that are based on human-defined rules (Rules-Based) in contrast to solutions that learn from experience (Intelligent).

We will use this framework to define and examine four categories of AI technology.

Business Intelligence.

Most modern organisations already use Business Intelligence (BI) solutions to generate interactive dashboards and reports. In essence, these are digital versions of paper-based reports, which we would not strictly classify as 'AI'. Traditional BI solutions need humans to define rules about the data to use, how to transform it, and how to present it - and they rely on people to do something with the information that they provide.

Be that as it may, the BI software companies are embedding machine learning technology into their products, evolving them into Advanced Analytics platforms (which we will cover in the next section). For example, Microsoft has added AI features to its Power BI software that allow it to understand questions in natural language and to generate a narrative summarising key insights from the data. Other BI software companies such as Qlik and Tableau are similarly upgrading their products.

If you need AI capability, we recommend looking at your existing BI software before acquiring a new product.

Advanced Analytics.

Advanced Analytics (AA) and BI solutions have a similar purpose - to inform decisions made by humans. However, AA solutions use machine learning algorithms to detect patterns in data that are hard for humans to find - including unstructured data such as audio, images and text. Applications of AA include:

• Predicting how customers with different characteristics will respond to marketing communications.


• Detecting anomalous sensor readings in equipment to recommend maintenance actions.


• Identifying objects through computer vision in order to assess risk, manage inventories, or trigger actions.

In addition to the evolving BI solutions we identified previously, there are specialised AA software tools. The most popular commercial options are the SAS suite, SPSS from IBM, and Matlab from MathWorks. These are enterprise-grade tools that increase the accessibility of AA to users who do not have advanced degrees in computer science or statistics.

While the commercial AA tools generally provide a better user experience than open source alternatives, increasingly organisations are supplementing or replacing them with free software such as R or machine learning libraries that they can use with general programming languages - Python being the most popular.

Robotic Process Automation.

In recent years, Robotic Process Automation (RPA) has emerged as a tool to automate simple tasks. RPA needs people to define rules for the software to follow so they lack the 'intelligence' part of AI. Nevertheless, RPA has been applied successfully and widely by large organisations to improve the efficiency of back-office functions such as Finance and HR.

The largest RPA software companies include UIPath, Automation Anywhere, and BluePrism. You can also find vendors that specialise in one or more business functions or industry verticals, such as Pega. Similarly to BI software, the RPA companies are adding AI capability to enable their products to learn rules from experience and work with unstructured data - they may refer to this capability as 'Intelligent Automation'.

If you need to connect one or more of your back-office systems, you might be able to use RPA to reliably copy data between them without making expensive software changes.

Cognitive Computing.

Cognitive Computing (CC) systems have both intelligence and the ability to take action. They are trained - based on example - to decide by themselves how to respond to specific events. Examples include:

• Marketing engines that send context-driven and time-sensitive offers to customers


• Virtual customer service agents


• Self-operating and self-optimising machines

Amazon, Google, IBM, and Microsoft offer AI services through their cloud platforms, which you can use to build your CC solution. This solution will typically include data, AI, and user interface modules to produce a complete application. The accuracy and speed of CC can vary significantly dependending on the services you use and how you integrate them.

The alternative to building your own CC solution is to buy one from a vendor that specialises in your niche. These vendors tend to be startup companies that build their product using cloud services. They can significantly accelerate your AI project, although they require more effort initially to evaluate their financial stability and compliance with industry regulations.

Since CC are autonomous, you will also need to implement operational and financial controls to manage them.

In summary.

We have examined four types of AI technology that have distinct business applications. We used a simple framework to define these categories, which you can also leverage to classify AI products based on autonomy and intelligence. You can analyse your organisation's AI portfolio by overlaying investments on top of this framework.

At Cognis, we are passionate about protecting the future of community-oriented organisations by enabling them to effectively engage with stakeholders in the digital economy.

What we do