The Future of Tech //
Hello, my name is Melody and I have just set up my own company working as a freelance programme manager for digital projects and products. I was asked by the lovely Nat at TWOP to write an article about the future technology. I thought I’d give you an introduction to Artificial Intelligence, as I’ve had some exposure to it in recent years, and having shifted from the ad world to the tech world I wanted to share some of my experiences around how it works, the paranoia, the positive impacts and what it could mean for our future.
So first up what is AI? Artificial intelligence (AI) is intelligence exhibited by machines. In computer science, an ideal “intelligent” machine is a flexible rational agent that perceives its environment and takes actions that maximize its chance of success at some goal.
When talking about the future of AI, everyone is very impressed (and often afraid) with the future vision of the technology. Reproducing human functions in digital form is something that many future facing companies are desperately trying to achieve. But fear not, it’s not all about Terminators and job stealing cyborgs.
Uber has recently rolled out a fleet of self driving cars which could help to prevent some of the thousands of deaths and casualties on the road each year. Google DeepMind is working towards solving intelligence (no mean feat!) and have made multiple huge breakthroughs in their research, training a program to beat a world champion in the Chinese strategic game of “Go“, supposedly one of the most difficult games in the world. They have also made a huge breakthrough teaching their programs to replicate human speech, reducing the gap to human performance by over 50% compared with the best competitor text-to-speech systems. On the surface it may seem fairly light touch with that kind of a mission statement but starting with strategic approaches to thinking is the perfect way to approach machine learning by combining logic with unique situational circumstances to achieve a single end goal. Teaching a machine to beat a human who has spent a lifetime to become an expert playing one of the world’s most renowned strategic games in the space of maybe one to two years gives you an indication of what humans and computers can achieve longer term using machine learning.
Replicating human speech and behaviour is something that has been already used commonly by Twitter and Apple’s Siri. People are already using this on a light touch level in more personalised ways with virtual PAs and call centres, although currently, I would always prefer to speak to a person.
Obviously this could be seen as a threat to the human based job market, or it could be an opportunity. If companies are guided by good morals rather than greed then this will mean they will have more money to put into intern and research programmes, meaning people could have greater access to funding and more people will be able to get into technical or creative pursuits. It all depends on how savvy our governments and tech leaders are to put particular laws in place that support the ever growing global population and the increased number of jobs that can be handled by robots.
Visual recognition by computers is something that companies like Facebook and Google are increasingly getting better at. Think about how many images you have ploughed into those platforms and tagged with data for them. There are already a number of public APIs that you can hook up to in order to visually recognise objects in an image or a scene. Not only can you recognise what something is but you can also get contextual information using emotion recognition, pose recognition and location detection to understand more about that particular scenario. And if you can understand from an image what someone is doing or where they are you can give them real time information about what’s going on around them or what they might be interested in, from telling people about local events, identifying what they’re wearing and using this to recommend local fashion boutiques or brands based on what they’re wearing. Or say hey, you’re in a crappy mood today, here’s a voucher for a free margarita – who doesn’t want that?
So already we live in a world where robots can be programmed to drive, talk, see, strategize, contextualise and, most importantly, learn. Which means that there is going to be a future wave of Artificial Intelligence that is more human in its nature and by humanising each of these features, this means that we can create more emotional AI and maybe a “robot race” that can collaborate with us to help us as humans do things better, solving global debt, curing famine and disease, or just sending us a free margarita when we’re in a bad mood – who knows.
The beauty of teaching a program to learn something is that you can program it to teach and re-teach itself millions of times within a short period of time. This reinforced learning approach is something that humans can only perform at a much slower rate. This was the approach used for the Alpha Go example by Google DeepMind and the same approach is used for many deep learning technologies.
When creating Artificial Intelligence machine learning clearly has the edge when it comes to reinforcement learning over humans. If you can write a program for it, then a machine will learn it faster than a human. But when it comes to more complex and individually unique things like emotion, judgement, morals, perception – how do you program that with all of those independent, organically grown, experience based variables and values. Earlier this year, Google created an Artificial Intelligence ethics board, the details of which remain much of a mystery but I would take the theories of big red self destruct buttons with a pinch of salt. I think this demonstrates the fact that they are conscious of the power of their technology and are willing to handle it responsibly.
AI and the robot revolution is predicted by many experts in this field to be a collaboration between humans and robots, maybe even creating cyborgs or working with humans to solve huge world problems like climate change, poverty and disease. In some cases there are things that humans will need robots for and a vice versa. But in developing these intelligence technologies, the tech giants working on these products have a social responsibility, not just to their own business, but of the potential political and economical impacts on our future world and they need to accept and address that they are potentially creating a new global superpower. Either way, we will see some huge exciting advances in this technology over the next decade or two and if you’re a part of it, then you have some influence over how this world will turn out.
Words by – Melody Michaud