AI - What's it all about? A simple guide.
We all know when to talk about AI… and that it’s changing the face of how we all work and live… but it can be hard - even for those working in the tech world - to know exactly what it is and the opportunities (and risks) AI presents.
In this simple guide we explain the basics, as well as some of the areas for consideration as we consider using AI in our businesses.
Back as far as the 1950s, the concept of AI was being discussed. The field was first explained as being “any task performed by a program or a machine that, if a human carried out the same activity, we would say the human had to apply intelligence to accomplish the task”.
The sorts of things AI systems demonstrate include planning, learning, reasoning, problem solving, knowledge representation, perception, motion, and manipulation and even social intelligence and creativity.
In today’s world we’ve got used to using AI even when we don’t realise it. Uses include using virtual assistants (like Alexa and Siri), recognising who is in a photo on Facebook, spotting possible spam in our inboxes, flagging up possible credit or online fraud and presenting adverts for purchases we might be interested in.
There are two forms of AI:
Narrow AI - intelligent systems that have been taught/learned to carry out specific tasks, without being programmed to do them. This includes organising personal calendars, responding to simple customer services queries, carrying out visual inspections on infrastructure and even interacting with other systems to do things like booking hotels, identifying wear and tear in equipment and flagging inappropriate content online.
General AI - the more flexible form of intelligence currently seen in the movies (like The Terminator, for example). Although this doesn’t really exist at the moment experts are generally of the opinion that it’s on its way… probably within the next 50 years!
So, how does ‘machine learning’ differ from ‘AI’? This feeds into the wider subject of AI and is a subset of AI research. It’s where a computer system is fed a large amount of data, which it uses to carry out a specific task. An example of this is being fed examples of different scans and reports on tumours, so that it can learn to more accurately detect them in the future.
This process works very much like a human brain. Neural networks in machine learning are interconnected layers of algorithms, called neurons, that feed data into each other. They can be trained to carry out specific tasks and to apportion importance to the data, as it is fed between them. During training of these neural networks, the weights attached to different inputs will continue to be varied until the output from the neural network is very close to what is desired, at which point the network will have 'learned' how to carry out a particular task.
This is expanded into what is known as ‘deep learning’ when these neural networks are expanded and expanded to be huge, with multiple layers, covering vast amounts of data.
There are different types of neural networks. In particular, recurrent neural networks are particularly well suited to language processing and speech recognition. Convolutional neural networks are more commonly used in image recognition.
Machine learning can be both supervised (where data is annotated to highlight the features of interest, allowing the machine to ‘learn’ to apply these labels to new data) and unsupervised (where algorithms look for similarities that allow them to identify patterns and categorise data themselves).
‘Evolutionary computation’ is another, evolving area or AI research. This works very much like Darwin’s theory of natural selection. Algorithms are subjected to random mutations and combinations, between generations, to evolve the optimal solution to a problem.
Finally, ‘expert systems’ are computers that have been programmed with rules that allow them to mimic the behaviour of an expert. An example of this is an autopilot system on an aeroplane.
The future of AI
The most obvious, expected, growth area is in the use of AI in robots and driverless cars. This stands to impact across the logistics, retail, transport and manufacturing sectors. Similarly, we can expect to see growth in the use of AI in technology relating to speech and language as well as facial recognition, surveillance and a significant growth in its use within healthcare. This latter may not only relate to more accurate diagnosis but also the recommendation of appropriate treatment and even the administration of such care.
It’s not all positive, however. In recent years there has also been a growth in the use of AI within the generation of ‘fake news’ and ‘technological warfare’. What does seem clear, however, is that AI is here to stay - whether that’s in reordering our milk and butter through the fridge, delivering our Amazon parcels or detecting poor health in a loved-one.
We’ve written a number of blogs about the legal challenges facing those working with AI. In particular these can include matters relating to Intellectual Property, data protection and privacy, as well as contractual terms. You can read more here: