Sitting in the audience of IBM’s excellent AI / IoT seminar this week, it occurred to me that AI has been talked about for a really long time, has been anticipated as ‘almost here’ for some years, and is upon us in a rush. In other words, it is following the usual path of technological development and adoption.
I like to refer to the three phases as:
- I dreamed
- I waited
- I blinked
Of the thousands of articles about Artificial Intelligence (AI) that have crossed my path over the last few years, I have read several dozen (some were very short), since it is a subject close to my heart, for several reasons.
My first serious look at the subject was more than 30 years ago when, as an undergraduate, I wrote a paper about the development and likely implementation of what were then called ‘Expert Systems’. The mainstream concept then was that you codified (literally) the expertise of leading professionals in a subject area (e.g.: property law, accounting), to develop what was, in essence, an interactive reference resource that a non-expert could reliably use, in place of that expert. Because the research at the time focused on replacing regulated or mission-critical professions, nothing much happened.
In the paper, I suggested that these ‘Expert Systems’ would be developed into powerful self-directed (not just interactive) databases that a qualified professional could use as a research / development tool, or a suitably-trained senior manager could use to replace routine tasks that require little professional input. My worked examples were producing legal documents and automating basic accounting, probably because these were subject I had studied and was familiar with. In both cases, replacing several junior staff, hugely increasing the productivity of senior professionals, was the outcome I envisioned.
Having been wrong for 30 years, I may be about to be right.
So, how did we get here, and where is AI going?
Well, let’s define ‘here’.
Expert Systems are among us. A good example is Zegal (the contract generator formerly known as Dragon Law). In client-facing Expert Systems, it is important that the user is not able to change the output (in this case, a contract), except through guided decision-making, or it’s no longer ‘Expert’.
Machine learning is growing by leaps and bounds, so it’s important to understand what this is (and is not). Basically, Machine Learning refers to a system’s ability to adjust responses in response to information, rather than have to be explicitly programmed (as is the case for Expert Systems). Amazon’s ‘Recommended for You’ function, Uber’s timing alerts, and Siri are well-known examples. The key difference between Machine Learning and Expert Systems is that the latter is fixed, and the former is responsive to data.
Chatbots and Virtual Assistants (currently focused on appointment setting) are great examples of Artificial Intelligence. AI enables natural language interaction, with responses based on semantic rather than simple key word identification. This is important, because meaningfulness of customer interaction, which drives engagement, which drives satisfaction. Because AI is that step ahead of Machine Learning, AI agents can seem almost human (the Virtual Assistant Evie is occasionally asked to attend events and sent gifts).
The concept of a machine-human entity (AKA Cyborg) has focused on the physical, but it is here in the virtual world, as many human agents are AI-assisted, with information and suggested responses being served to them by the AI agent. The next time someone has a complex answer ready for you in a suspiciously short time, you will know you are dealing with a Virtual Cyborg.
So what is the future of AI?
I will come back to this in a future article, but a strong current trend is for companies to differentiate between high-volume and high-touch contacts, to automate the former and AI-support the latter. As this requires far more data than is currently collected, expect organisations to convert even phone conversations to data…but that’s a whole new conversation.