One key element that will impact all our futures is artificial intelligence. Now, I have written about AI on several previous occasions, and I also feature an entire chapter on its applications to learning and development in my new book, Digital Learning in Organisations. But here, on this more informal channel that I call my blog, I just wanted to share with you a few thoughts about the impact I think it will (and won’t) have on our learning future. The term ‘artificial intelligence’ is often misused, because many of us fail to recognise that there are many levels of AI and misunderstandings can occur. It’s important that we distinguish between the several kinds of AI and recognise their differences in a critical manner. Here are some thoughts:

Firstly, AI has been overhyped. There are many ‘in the know’ who claim that AI will radically transform our lives. For some that may be true, but the stark reality is that even if AI reaches its full potential, it will be for those in the developed world, while hundreds of millions of other citizens of this planet will continue to live out their lives unaware of its impact. The digital divide between those who have access to technology and those who don’t, continues to grow wider and more pronounced as technology advances. As writer William Gibson put it ‘The future is here. It’s just not evenly distributed.’ Bottom line: AI will only benefit those who live in the connected regions of our society.

Secondly, there are some who claim that AI is an existential threat to humanity. At the highest level of AI, known as Artificial Super Intelligence (ASI) this might be true, but only as physicist Max Tegmark suggests, if the ‘intentions’ of a super intelligent machine do not align to those of humanity. But the reality is that we are still working predominantly at a very low, ‘or weak’ level of AI known as Artificial Narrow Intelligence (ANI) where in many cases, AI has not surpassed human capabilities, and possibly won’t for some time. To reach the intermediate level of AI, which we refer to as Artificial General Intelligence (AGI), computers must first pass the Turing Test. This is a test created by mathematician Alan Turing in the 1950s to determine whether a machine could be sufficiently ‘intelligent’ to fool a human into thinking they were communicating with another human. Many would argue that this test has yet to be passed by ‘artificially intelligent’ technologies. Note that in the diagram above, the only AGI and ASI examples are from science fiction. IBM’s Deep Blue and Watson are specialist technologies, but limited to one narrow activity, i.e. playing chess or winning TV quiz shows.

Thirdly, AI is developing and is still nascent. The idea that a machine can perform the roles and duties of a human being has been with us for a long time. Popular novels and movies have featured numerous ‘artificial life forms’ or sentient machines such as robots that can match the performance of humans, if not exceed them. Although in some specific fields of activity AI can indeed best a human being, it is far from universal. At present most of the AI we see around us is at the ‘narrow’ level, meaning that computers are following code that instructs them to perform a single task. Examples of narrow intelligence include the supermarket checkout robots, personal virtual assistants such as Alexa and Siri, and internet chatbots. More advanced forms of AI that perform generally are rare, and expensive, and certainly not yet available for use by the general public.

Most futurists predict that things will change exponentially in the coming years. Computer technology will advance sufficiently to challenge all of us, but until it does, we are left with artificial (machine) intelligence that is limited in scope, but that is becoming increasingly indispensable.

Previous posts in the series ‘Our Digital Future’:
1: Gazing down the corridor
2: Smart clothing
3: The semantic web
4: Pervasive computing

Our digital future 5: by Steve Wheeler was written in Plymouth, England and is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.