Welcome to Antenna, our glance into the middle distance and the life-changing developments the near future will bring. We’ll also look at the investment implications which, as ever, are far from intuitive.
What is AI?
Scratch the surface of this question and you will find layers of different technologies and fields of study. But fundamentally, it is the development of machines that behave like human brains, performing tasks that would otherwise require human intelligence, such as learning, planning, reasoning and problem-solving. Although that might sound futuristic, AI is already here. It powers your internet searches, enables your voice-activated personal assistants, recommends products online and detects fraud on bank accounts, among many other things. Combined with other technologies, it also enables robots and driverless cars.
All AI that exists today is classified as narrow (sometimes called ‘weak’), which means it can perform only the specific task it is designed for. Technologists are working towards artificial general intelligence (AGI, or ‘strong AI’), which would perform any intellectual task a human can.
How advanced is the technology?
We are still a long way from creating AGI, but we are getting closer and the journey is being driven by computer games such as chess and Go. UK-based AI company DeepMind, owned by Google, seems to be leading the way with its ‘Alpha’ series of programs. AlphaGo Zero effectively became the world champion at the Chinese game Go in 2017. Two things make this remarkable. First, it is impossible to win the game simply using the force of processing power because there are more possible moves in Go than there are atoms in the observable universe. Second, unlike previous models, the program trained without examples of human play, using just the rules of the game.
When machines teach themselves in this way, they are not constrained by human knowledge and think in what DeepMind co-founder Demis Hassabis has described as an “alien” way. At a conference shortly after the release of the company’s new headline-grabbing program, AlphaZero – which builds on AlphaGo Zero with the ability to learn not just Go, but also chess and Shogi (learning more than one task being a major hurdle in developing AGI) – Hassabis said: “It doesn’t play like a human, and it doesn’t play like a program.” He described maverick chess plays such as moving the queen into the corner of the board and reflected: “Maybe our conception of chess has been too limited.” The teacher, then, may become the pupil, as we are able to learn new methods from AI.
The ‘Alphas’ mentioned above are examples of machine learning (ML), or deep learning (a more advanced form), which is a subset of AI whereby systems learn from data without being explicitly programmed and make predictions. ML is thought to be the best path to AGI and is already in common use in things like internet searches. It is enabled by technology called neural networks. These comprise thousands of processing nodes, or layers of computation (think spreadsheets), that mimic the human brain by spotting patterns in images, sound and text. That ability to interpret the world is enabled by advances in natural language processing (NLP, see below) and visual reasoning.
So AI is a discipline that comprises many layers of research, a lot of data and an ever-increasing amount of processing power. When those ingredients are brought together with a pinch of human imagination and other technologies, practical applications are born. While it is impossible to predict all the uses humans will find for AI, the problems researchers are working to solve at the moment provide clues as to the direction we are heading in.
What will AI be able to do in the future?
NLP aims to reduce the gap between what a human says and what a machine understands. Siri and Alexa are sparkling examples of how far this technology has come. But NLP systems are in their infancy. They are improving steadily in their understanding of context and analogy, which is implicit to humans – for example, they are starting to analyse the form text takes, not just its constituent words, which reveals extra meaning (an abrupt email could signify an angry author). One area where NLP is already excelling is unlocking meaning from unstructured data, such as vast volumes of emails. High-street bank RBS uses NLP to spot trends in customer feedback by analysing emails and telephone calls. A related development is affective computing, which aims to create AI that can recognise and interpret human emotions and, ultimately, simulate empathy. Affective computing is applied to speech and video and is being trained in cultural context.
While AI may have better memory than humans, it is a long way behind us in its ability to reason – a core facet of intelligence. A human, for example, knows that if you stack too many boxes on top of each other, the bottom one will collapse – this is implicit. A machine needs to be told. Understanding the relationship between things and reasoning without specific previous examples is the cutting edge of research. Advances in this field could get AI systems thinking creatively about problems in the world around us, such as climate change, without the limitations of human understanding.
Step out into the world
The most obvious examples are robots that run AI programs, enabling them to traverse uneven surfaces, learning to balance as they go, and autonomous cars that interpret the world around them and take decisions accordingly. But our imaginations are the limit for how AI might get out into the world and transform our lives. One cutting-edge example that comes from Massachusetts Institute of Technology is an ingestible robot.
As the technology advances, the hope is that AI will become capable of adapting to different tasks and situations, rather than performing just the task it is designed for. Intuitive interfaces will enable users to be instructive with AI, so that eventually they will effectively be able to reprogram systems to do different jobs simply by talking to them.
Will the robots take over the world?
With researchers determined to build AGI, there are concerns that, at some point, machine intelligence could surpass our own. This is called superintelligence and its advent is referred to as the singularity, which, it is hypothesised, would trigger runaway technological growth and unfathomable changes to civilisation. Although we are unlikely to create a superintelligence, as we do not even understand how the human mind works, leading thinkers are keen for AI to be developed within a framework that will limit the possibility: they want safe AI.
IMPORTANT NOTE: The value of your investments may go down as well as up. Past performance is not a guide to future performance. Any tax allowances or thresholds mentioned are based on personal circumstances and current legislation, which are subject to change. Some products or services may be affected by changes in currency exchange rates. If you invest in currencies other than your own, the value of your investment may move independently of the underlying asset. All information within this publication is for illustrative purposes only and is not intended as investment advice; no investment is suitable in all cases and if you have any doubts as to an investment’s suitability then you should contact us or your financial adviser.
We or a connected person may have positions in or options on the securities mentioned herein or may buy, sell or offer to make a purchase or sale of such securities from time to time. In addition we reserve the right to act as principal or agent with regard to the sale or purchase of any security mentioned in this document. For further information, please refer to our conflicts policy, which is available on request or can be accessed via our website at www.brewin.co.uk. The opinions expressed in this publication are not necessarily the views held throughout the Brewin Dolphin Group. No Director, representative or employee of the Brewin Dolphin Group accepts liability for any direct or consequential loss arising from the use of this document or its contents. The information contained in this publication is believed to be reliable and accurate, but without further investigation cannot be warranted as to accuracy or completeness.