AI confusion reigns, but what are the real possibilities?
AI coupled to big data, is already causing disruption to some industries. Obvious examples are the way Airbnb and Uber have changed the market dynamics for short-term rental accommodation and taxi hire.
However, most people have no idea how the underlying systems for such new applications actually work. What will be disrupted next? And how do we all take advantage of these new technologies? What are the bounds for the next thirty years?
The really interesting thing is that the underlying processes are very simple, but the effect of joining up simple technology can provide massive opportunities (and disruption). The clever bit of AI coupled to big data see’s these fairly stupid computers doing seemingly quite clever stuff. This is of course a mirage, as none of the current technology can do much more than make simple comparisons and choose a relevant path from learning about what has worked before.
AI vs Canine
AI is much like your dog – you may train your dog to fetch the paper after it drops through the letter box. This may take some time. But while it seems to understand that you wish to read the paper as soon as it arrives, your dog has no understanding of what the paper is, or what you gain from it. It has just learnt to follow a process. No doubt these skills are somewhat transferable, change the reader along with the house and the dog is likely to apply the same procedure in new circumstances. Ask the dog what the paper is for and you will not be surprised at the answer.
Now, the current AI cleverness arrives from having a training system that uses millions and millions of bits of ‘test’ data. Today’s computers are now fast enough to be trained using such massive data sets quite quickly and this speed is increasing due to new types of computer chips that make such calculations in parallel, (rather than traditionally one bit of data at a time). This is somewhat analogous to how our brain works – we have many billions of neurons that work simultaneously to work out problems – we don’t actually know how we do this – but clearly it works!
The modern AI and big data computers use similar, but much, much smaller neural networks. The computer learns and re-learns from the test data and real data, so that it gets better and better at solving a particular test or set of circumstances.
So while these AI systems are quite good at face recognition, working out the best way to play a computer game or finding their way around obstacles in their path, they are still hopeless if you give them a task that they have not come across before. So, by training on specific tasks, your robot-dog can fetch your paper too. And while your dog probably has some inkling that you like your paper, the robot-dog simply has no idea about your existence or your paper.
The Media & Miss-information
The media have a lot to answer for in misrepresenting how these AI systems work. They often portray robots doing stuff that seems intelligent, the way they move, what they utter and how they respond can be beguiling. But in reality, there are no ideas or thought processes as we know it – just a learnt response to a set of data that the robots and AI system have seen many times before.
Now, of course, as we do not understand the brain, we cannot know when the AI system will acquire the complexity to become sentient. The size of current AI systems compared to the brain is tiny – so until we design a hugely more sophisticated device, this potential sentience is someway off.
But, importantly for the future of the human race, we can utilise the cleverness of the current and rapidly improving AI systems to do very many useful things. Some people believe that we have a limited timescale to improve the world before we destroy it entirely. This cleverness is now just about good enough to help humans achieve the aim of finding a new way forward.
Many things people do every day are no more than learnt responses to recurring actions and facts. This is especially true of repetitive jobs. It is also true, (although not so obvious), of people who work with large ‘data sets’. We know that AI is excellent with these – better than us – so knowledge workers such as accountants, legal experts and medical experts can be helped or potentially replaced by AI systems. This is already happening, IBM have their Watson computer helping researchers to find new drugs from past experimental data and helping doctors with complex diagnosis.
We also know about the autonomous cars and their steady improvement, Tesla cars and some Mercedes can already drive you, with your hands-off the wheel, down the motorway or dual carriageway. Google’s Waymo car is starting a full taxi service in Phoenix Arizona in 2018 – with no one at all in the front seats!
What happens when we join all these technologies up?
Now, what people miss is that if we join-up these technologies – the AI, big data, robotics, the sensors linked by fast wireless communications – then we have something very special in store.
All these technologies exist now. Imagine the new robot factory, the AI system is trained to make a new robot – a repetitive task, and the new AI system can pass on its learning to the new robot. This in turn makes another robot, with of course the help from automated manufacturing systems and automated supply, delivered to the factory by driverless trucks. It does not take a great leap of imagination to realise that we could, quite soon, have more robots than we know what to do with.
I have worked for many years with accounting systems. These are long overdue for an overhaul, and when this starts it will shake up the world. Accounting is very simple; we are just moving data into the relevant boxes. While big companies will struggle to change, smaller organisations will lead the way. There is no point in having people input account data, re-arranging it or pass it on to a separate organisation. Once the joined-up applications arrive to do this – any organisation that uses this new model will simply by-pass all the costs and drudgery of double entry accounting. New fintech technology such as blockchain will underpin the validity of all transaction to a more consistent level than possible today. Bye, bye accountants.
A new Vision for the Future
So, with a bit of vision, we could within a short timescale, perhaps no more than fifteen to thirty years, gain massive robotic production with a bit of cleverness thrown in, to drive us about, distribute things for us, make things for us, build and run the power stations, farm the fields, do all the accounting and banking, provide the best medical advice, the best legal advice, build most of our houses, and make the tea.
This means that humans could be freed up to different stuff, like solve the world’s most pressing problems. This brings us directly to a misunderstanding about wages, money and resources. We have all ‘lived to work’, and therefore have difficulty in standing back and seeing this future for what it really is. There will be no lack of ‘wealth’ as the robotics and AI systems will be doing a great job – effectively for nothing – we shall live in a land of plenty.
Money ceases to be an issue – it is the share of physical resources that becomes the central issue. If we are sensible we shall recognise that no one needs to own the robots, they should be a resource for the whole of humanity. How we adapt to this challenge may well usher in a golden era for the human race – or, usher in a new wave of global repression. Your choice!
Curious how we can upgrade our economic system so it supports the quality of your life and of those you care for?
In the book “Happonomy, Roadmap to Utopia”, Bruno Delepierre takes you on a 300 page journey to explore how work, money and technology impacts the quality of your life. Expect insightful analyses, intimate portraits and 35 daring recipes for upgrade. Interested? Take a look!
Additional AI and robotics resources: