by Kelvin Chapman
Artificial Intelligence – Challenge or Opportunity?
It’s the latest thing in the tech world that everyone is talking about. It promises much and seems to be delivering at least some of the promises. It is becoming more widely used but what of the doomsday predictions that are being made.
So what is AI? Is it the breakthrough that will allow many of the tasks that we now do to be done by computers perhaps executed by robots? And will these robots be able to make their own decisions – perhaps being able to turn against their creators?
Is it here now? And the answer to that is right now on our cell phones – every time you use the predictive text you are using an early form of artificial intelligence – one that learns from the messages we have previously sent – and from the message that we are replying to
To begin to understand all this we need to go back to the earliest computers – and perhaps a little earlier than that.
Alan Turing was the man behind the UK code breaking effort in WW 2 and as part of that work developed the first stored program computers – the earliest form of those we have today. These were comparatively crude. Transistors had not yet been invented. But Turing saw the potential of these new computing machines and in 1952 was asking the question “can automatic calculating machines be said to think?” This resulted in what has become known as the Turing test which in its simplest form postulates that if a human is communicating remotely with something and, whatever questions may be asked, cannot determine whether that thing behind the wall is a human or a machine. If it is a machine, that machine can be said to have the capability of the human brain.
We will come back to the Turing test later on in this discussion
The earliest computers could only act on the information they had been given. Their main benefit was that they could do things much faster than human computers.
But as computers have become faster the internet has come into being allowing us to access a vast range of information from our cell phones. Artificial intelligence, first defined by Alan Turing, is the term now used to describe computers and programmes that allow questions to answered using this broad range of information and to draw broader conclusions from the data available
These answers are more likely to be accurate where the rules that apply to the question are well established. One such case is where it is now possible to look for relationships in the health field, identifying relationships in data that are not identified by previous methods.
We are beginning to see previously unidentified responses – both favourable and unfavourable – to drugs in this data. AI has also enabled the process by which the proteins in all living cells can “fold” – an important next step in understanding how living cells replicate. These applications will result in a better understanding of how drugs work and lead to improve health outcomes far more quickly than earlier methods of analysis.
Another case is in the legal system where cases are fairly well documented to a common standard, making it possible to identify relevant details in past cases both quickly and comprehensively. Answering exam questions is a doddle for an AI system where the course material is available on the internet.
And of course AI can correct our grammar much more accurately than we perhaps would like, while also identifying ways in which text can be made clearer.
These are useful and positive applications. The same rules enable an AI system to take a voice sample and a face photo and create a video of the person behind the face saying anything that the author wants. The possibility for deception is very real – and is with us now. Continue reading Sunday 21st July