The term “artificial intelligence” resembles the term “blockchain” in that while you have undoubtedly already heard, read and maybe even talked about it – you may still not be entirely sure what it actually means.
In the first of two workshops forming part of our Specialist in AI for Business certification course, we give participants a solid introduction to the main characteristics of artificial intelligence. In terms of subject matter, we focus on trends, latest developments, functions and current areas of application in a business context. We also discuss how AI actually “learns”, given that artificially intelligent machines still depend on human input.
At present, the technology still relies on predictions. While this word positively encourages you to imagine the Delphic oracle, in actual fact it has very little to do with it. While the term “prediction” is derived from the Latin “praedicere”, meaning “to preordain” or “prophesy”, in this case we should interpret it as meaning “to deduce” or “extrapolate”. As things stand, AI still works by gathering together large amounts of data and then working out what information is missing. Thus you generate the information you need based on the information you have. This suddenly sounds a lot less magical; surely this is the kind of thing human beings can do? Well yes, they can – but much more slowly and much less reliably.
Google Maps is a good example of AI’s predictive precision. Google Maps can do anything which any human familiar with maps and the local environment could also do, such as guide you from your hotel to the airport. But Google Maps gives you the information you need many times faster than a human could. In addition, you can also use the program’s extra services at any time, where a human being might not be able to oblige. And Google Maps is also capable of doing other things which would be very difficult for most people to do, such as assessing current traffic levels, planning the best or shortest route (in real time) and quickly finding its bearings again if you happen to make a wrong turn.
Another real-world example illustrates how unspectacular-sounding deductions can actually achieve a great deal. In its former incarnation, Google Translate used a purely grammar-based approach. Sentences were translated by applying grammatical rules, meaning that the output often sounded like gobbledygook. To solve this problem, members of the Google Brain Team (yes, it really is called that!) partially redesigned the system so that it no longer translates solely by applying grammatical rules, but now uses much the same approach as a human translator. Google effectively used prediction to master the translation challenge.
Is prediction the same thing as intelligence?
Many readers will undoubtedly be wondering whether you can really call prediction an act of intelligence. Yes, you definitely can! But opinions differ on the extent to which this is true. The fact remains that deducing correlations (interrelationships) is a function of intelligence.
Jeff Hawkins, U.S. IT entrepreneur and neuroscientist, goes one step further, claiming that: “Prediction is not just one of the things your brain does. It is the primary function of the neocortex and the foundation of intelligence.”
What’s more, it is possible to construct a prediction. By continuously multiplying data values, artificial intelligence is capable of learning and improving its abilities (a process called machine learning).
Exactly how artificial intelligence can be technically implemented in practice is something you discover (along with plenty of other things) in our second Specialist in AI for Business workshop. You can find full details of the course on our website.