Future of Artificial Intelligence. How AI will change jobs, manufacturing and health. Impact of AI on marketing, investment, banking, insurance, politics, government, warfare, consumer choices and how people think. Artificial Intelligence FEATURE

Futurist Keynote Speaker: Posts, Slides, Videos - AI Artificial Intelligence Robotics Cybersecurity

Artificial Intelligence will completely dominate and transform our world in ways that very few people understand today.  

And it's all happening very rapidly - whether robot drivers, Artificial Intelligence health care advisors, or drones owned by the security services which are controlled by AI, with permissions to kill automatically. 

The big question is what it all means for us all, in the way we live now or in the future?  Could AI go out of control?  Will robots ever achieve full human consciousness, or "real" intelligence, or will they just remain smart machines?

Here is a lengthy post by Keynote Speaker Patrick Dixon, based on lectures to hundreds of multinationals on AI and related issues.

Robot drivers – Artificial Intelligence licensed to kill

One of the greatest investments in AI today is in robot drivers of cars, vans, lorries and buses. We can debate about the speed of uptake of Artificial Intelligence and robotics, but not about the actual trend which is beyond doubt.

We are about to see a massive jump in the power of AI driving machines for example.

Imagine a world where all vehicles are supplied with road sensors, cameras, detailed mapping capability and continuous online communications, as part of the standard purchase price. A world where in-car computers, all linked together across the world, are learning from every second on the road:

- from every twitch of the wheel by every human driver

- from every touch of feet on brakes or accelerators

- from every human decision made about the road conditions, and from every accident

The first robot drivers were trained on specially equipped cars, and rapidly became smart enough to start driving on their own, without human supervision.

But now we are talking about shared experiences from hundreds of millions of drivers, all collected and analysed automatically.

The result will be that robots will be far cheaper to unsure than humans.  

As you get into the car you will select manual or robot, and if you switch to manual, you will receive a warning from the insurer that the cost per mile when you drive yourself will be up to twice as much as the cost if the robot takes over, because they have far fewer accidents.  And probably you will also hear or see a safety warning for the same reason.

Robot drivers have to be able to make complex ethical decisions. No robot can take to the roads without being given a licence to kill.

The reason is that just like human drivers, robot drivers can easily be confronted with split second life and death decisions:

Two children run out suddenly into the road chasing a football, right into the path of an oncoming vehicle.

The braking distance is too short to save the children from impact. Their only chance is for the driver to swerve violently to the left or right. To the left there is a pavement where a mother and three small children are walking along. To the right on the other side of the road is an old man standing with a dog.

What is the driver to do? What is the robot to do? Artificial Intelligence will make a decision in nanoseconds after calculating various risks, guided also by an "ethical scorecard". 

But will it be the "right" decision?  Could a super-powerful AI start to change its own rules?

Most Artificial Intelligence systems still quite limited in focus

For the most part, all what we are seeing are clever pieces of AI software, crunching large amounts of data, doing with it precisely what AI programmers determined in advance.

We are NOT seeing many examples of computers writing new computer code for their own brains. We are NOT seeing true leaps of insight.  We are NOT seeing anything approaching the sophistication of human consciousness.

Common sense depends on more than lines of computer code – however cleverly those lines are written.

However, Machine Learning is taking AI a step further. Here the software code itself has the ability to deepen its own understanding. 

So it means for example, that a particular robot can become expert in playing a game, just by watching people, without even being taught the basic rules.

This is fascinating.  It leads to the possibility that a robot could teach itself to understand human speech, without knowing anything about language.

But the same robot will probably struggle to understand why a human is feeling sad.

And human consciousness is a huge step beyond that.

Health care – Artificial Intelligence diagnosis better than doctors 

Artificial intelligence is already having a major impact on health care. This is especially the case in making a diagnosis and deciding on treatment.

Babylon is just one example of an Artificial Intelligence system to make decisions about patient care.

Algorithms have been used in medical call centres for decades, to provide after hours health care advice in emergencies, decide whether to call an ambulance and so on. These are simple flow charts. Each answer to a previous question is used to trigger the next question, until a decision-point is reached. These guides for nurses and doctors have undoubtedly saved huge numbers of lives.

Babylon is a mobile App which automates this process for people seeking medical advice. Instead of calling a helpline, and answering questions over the phone to a human being, patients write their own answers into the App, which then makes a decision about what to do. The App can trigger an immediate ambulance, a call back by a nurse or doctor, tell the person to visit a pharmacy, or just provide reassurance.

These diagnostic systems are improving all the time, so long as outcomes are fed into the system so that there is constant feedback about what actually happened later on, and what the confirmed diagnosis turned out to be.

Babylon is already more accurate than most doctors or nurses working on their own with only the normal call-centre flow charts.

Similar diagnostic and decision-making Artificial Intelligence tools have been used in Intensive Care for many years, and can be more accurate than even very experienced physicians.

AI therefore has radical implications for patient care, medico-legal cases and medical insurance cover. Take for example a patient in intensive care where either the doctor decided not to consult the AI advisor, or decided to ignore the AI advice.

If the patient dies, and relatives sue the hospital, the prosecution lawyers may have a strong case for negligence. The defence lawyers may argue that the patient was very sick and present expert opinion that the treatment plan was a reasonable one. But juries may well be swayed by research showing that on average the AI advisor has a better track record, has more data in its system than in the brain of any doctor, and is therefore more competent in most circumstances to make such decisions.

So we then we will end up with a situation where power has been stripped away from very experienced physicians, reduced to mere servants of the medical robot. Further into the future we can expect big debates about what kind of medical training is needed, what knowledge is required, in a health care system where expert medical knowledge and experience of human beings is usually over-ridden by even more expert AI systems.

Can Artificial Intelligence replace a psychotherapist?

Back in the 1950s, the famous computer genius Alan Turing posed the so-called Turing Test for true Artificial Intelligence, which was that a human being should not be able to tell if they are interacting with a human being or with an AI machine.

It's a very important test and many thousands of attempts have been made to match this AI standard. 

One of the most interesting early tests was to try to simulate a pyschotherapist. I remember playing with many such interfaces back in the early 1980s.

The person sits in front of a screen and begins to answer AI questions such as:

"How are you feeling in yourself today?"

The most convincing software would make links between one statement and another. For example, by replying:

"Do you think you are not sleeping at night because you are angry with your mother about your childhood?"

A great test is to put a real psychiatrist in one room, and the person in the other, linked by a chat screen, and at a particular point allow the AI text generator to take over, before swopping back again. The aim is to see how many times you can swop back and forth without the person being able to work out who is who.

Actually it is relatively easy to create an AI for psychotherapy, because of the structured way in which many first interviews are carried out.

But such systems, while clever, are really very limited.  Human beings have mapped out large numbers of responses, or rules for responding, so that the software produces answers that feel human.  But it's just software at the end of the day.  Yes, the smarter software allows data to be gathered and patterns recognised, so that new responses are formed which are not words or phrases created in advance by a human, but again, using various language rules that human beings have devised.

AI impact on finance, economics and investment decisions

AI is already having a profound impact on stock markets. The great majority of all trades are conducted by robots, operating automatically at the speed of light, interpreting vast ranges of data from around the world. These robots are constantly scanning news feeds, social media, weather forecasts, economic data from government agencies, and of course all buying and selling by other robots as well as people.  But once again, lacking true intelligence.  These robots are programmed to detect patterns, and to trigger certain actions in response - again by humans.

This can produce some very strange results. Because robots are watching decisions by other robots, and because all their decisions to buy or sell are largely pre-programmed by software engineers working with investment experts, their combined efforts can over-react, with wild selling or buying.

And its all to do with speed. When tens of thousands of AI systems are making investment decisions based on the same kinds of human rules, based on similar sources of data, what really counts is getting trades completed just a few milliseconds before all the other Artificial Intelligence platforms do so.

That's why these AI robots have to be located very close to where markets are traded. It's also why the owners of such AI invest huge amounts of money in ultra-fast processing and data transmission.

So you can see that human beings are being completely out-gunned by the collective fire-power of investment AI.

Many large investment funds are managed entirely by Artificial Intelligence. These "passive" funds are known as Trackers, and are set up to follow major stock market indexes automatically.

If a company is listed in the top 100 stocks on the London Stock Exchange for example, then AI will automatically buy its stock, ensuring that the amount of stock held is in proportion to the total value of the company compared to the entire value of the top 100 companies.

Research shows that in up to 70% of cases, Artificial Intelligence investing produces better investment returns in passive tracker funds than expensive fund managers, when you take all the costs of those fund managers into account. These passive funds cost almost nothing to run.

Fraud detection by Artificial Intelligence

AI is also being used by banks to detect fraud. The most common way this is done is by detection of simple variations in normal patterns. For example, you may live in the UK and sometimes go on holiday to Europe. You would not usually buy things on a credit card in Brazil or Nigeria. So the moment the bank detects transactions from such nations, an alert is triggered which may lock the card immediately, as well as a message to you to contact the anti-fraud team at the bank.

But the largest frauds are more subtle, and often involve staff working away to beat control systems. A single major fraud can bring down an entire bank, if huge investment positions have been taken in complex ways, resulting in gigantic losses, which maybe continue to grow over time.

AI is being used to detect major frauds like this before they happen. Artificial Ingtelligence is able to sense very subtle but slightly unusual patterns – maybe phone calls from a trader to an unknown number at strange times of day; emails which contain certain words; unusual clusters or frequency of trades, and so on.

Underwriting risks in insurance using Artificial Intelligence

The insurance industry is also investing heavily in Artificial Intelligence, to gain a better understanding of risks, set premiums more accurately, and detect claims that are likely to be fraudulent.

This is especially a problem with car insurance, related to claims for things like whiplash injuries. AI systems are being used to cross-correlate different factors – for example to try to pick up if multiple claims are coming from a particular family, based in different locations, or from different occupants of the same block of flats and so on, as such false claims are often organised by syndicates or gangs.

Predicting equipment failure using Artificial Intelligence

Companies like Infosys are using Artificial Intelligence to predict major industrial disasters and equipment failures before they happen. Infosys has installed large numbers of sensors on oil rigs, and rigged them up to Artificial Intelligence machines, which start to learn from patterns.

After some time, AI started to generate alerts, based on patterns of noise, vibration, temperature, weather, sea state and other variables, which were often very specific about which part of the rig was about to blow up or which pump was about to fail.

The same is now happening with transatlantic cables for telco companies. AI is able to detect early warning signs of trouble before data flow is compromised significantly, to enable rapid repairs.

Predicting sports results using Artificial Intelligence

Infosys has also taken a lead in predicting sports results – with unnerving consequencies for players, coaches and major new opportunities for gamblers if they can get hold of the AI forecasts themselves.

Infosys started feeding AI robots over 40 years of video of professional tennis matches from major competitions all over the world. They then added in data about the speed of every ball served, where each ball landed, and so on, at every stage of every game, analysed differently for every player.

The Infosys team were astonished to find that Artificial Intelligence was soon able to predict more accurately than anyone would have guessed, where a particular player was most likely to place his or her next serve, into what part of the court, at what speed.

They were able to show that even the best professional players are far more predictable in how they behave on court, than their opponents or their coaches would ever guess. Such insights are invaluable in studying how best to win against a particular player, or how to deal with particular weaknesses in one's own capability.

Social marketing, citizen tracking and thought-detection using AI

Artificial Intelligence in marketing is becoming a major driver – not only in online / digital / mobile marketing, but also in niche targeting of more traditional campaigns such as mailshots, cable TV adverts and tele-sales.

Recent research shows that the answer to a single question on Facebook can predict with 40% accuracy what the response will be to a wide range of marketing offers. The big question is what is that question!

And of course the answer is that it all depends. But the point is that Big Data analytics means we can compare profiles and other information in facebook feeds with what items people like, what posts they make, what sites they click through to, what videos they watch, and the things they actually buy.

Semantic search based on Artificial Intelligence

Linked to this is the explosion of semantic search. Ten years ago, search engines like Google or Yahoo tended to use the words entered into a search request in quite a rigid way. If you typed in "Keynote speaker on AI", the engines would probably ignore or push down the listings any web pages that referred only to "conference lectures" and "artificial intelligence".

Today, if you type in "AI speaker", Google is likely to be able to work out that you are probably a conference organizer looking for keynote speakers on Artificial Intelligence, robotics, automation, Big Data and a range of connected themes.

And the moment you start clicking on items Google lists, you are feeding extra information to improve the next search request. More than that, Google will also wait to see how long you linger on each page you click on, to learn even more about what your searching is really about, what is in your mind right now.

Of course, Google can also add insights from scanning the content of the last few thousand emails you have sent and received on gmail, data from videos you watch on YouTube, the posts you make on LinkedIn or Twitter, information from your diary appointments, and data about your current location as well as where you have been travelling to recently.

All this adds up to an extraordinarily detailed picture of who you are, but this kind of Artificial Intelligence has hardly begun to develop yet. We can go much further.

Photo mapping and face matching using Artificial Intelligence

One of the most powerful parts of the total AI universe today is the rapidly growing ability of machines to recognize people from a distance – either in still photos or in video streams or in recordings of their voices.

These technologies are improving in an astonishing way. We are seeing this on both iPhone and Android phones, where AI is scanning entire collections of your personal photos, picking out individuals it recognizes in other photos. And even if you fail to enter the name yourself of the person on any photo, AI may often work it out independently from other clues – for example a public Facebook profile page, or Twitter profile or LinkedIn profile, or from some post from someone else where the person has been tagged.

And that is before we link national AI systems up with driving licence databases, passport databases, travel card indexes, storecards, workplace identity cards, and a host of other sources of information which contain your name and your photo.

The question is how to use such data without freaking out your own customers. I was recently at a conference giving a keynote on digital innovation including AI. In the exhibition area was a mock up of a supermarket checkout.

Hidden away was a very small camera, which was instantly recognizing each customer, so that the chain could bring up a specially customized offer on the screen at the moment they were about to pay.

Airports are using AI face-matching to track tens of millions of travellers a year as they walk through security, wander around the shopping areas and make their way to the gates or board planes.

At another conference, I walked up to a camera which was scanning faces of people who walked by. The AI system reported accurately my age, gender and mood – which it monitored accurately right in front of me, from surprise to delight to slight concern.

Surveillance and censorship - abuse of Artificial Intelligence?

So then, without anything like secretly turning on phone microphones or computer cameras, the state already has very powerful tools to track and monitor people using Artificial Intelligence.

Especially when we combine face-matching with all the data generated by mobile phone companies, which is constantly reporting the position of all mobile phone users within a metre or two. On most handsets, your location data is being recorded even when the phone is turned off, out of signal range, even if there is no WiFi or any other electronic connection of any kind.

To this we can add the hundreds of ways that exist to turn just about any type of mobile device or computer into a bugging tool, using simple software hacks, new types of which are being discovered every week.

The usual destination for these discoveries is into the secret archives of national security agencies, where they are further developed as exploits to monitor people of interest. So the computer or phone companies themselves are often left completely unaware of hacking vulnerabilities, maybe for many years.

That's why it was such a disaster, and so embarrassing when a collection of around 100 sophisticated hacking tools assembled by the US National Security Agency (NSA) was leaked online in March 2017. As soon as this happened, criminals and hostile states were locked in a race against time to exploit as many of these as possible before patches were designed and published by the IT companies affected so that everyone could upgrade their security on their phones or computers.

Of course, many people have downloaded Apps or use devices that specifically give permission to "bug" conversations or background noise all the time. Take for example Siri or Amazon Echo. These devices are constantly listening for key words to activate their command centres, so the technology is already many steps towards the perfect bugging solution for any state operator or a criminal.

All you need to do is trick the device into transmitting continuous sound even when the command such as "Hey Siri" has not actually been said.

Another example is Shazam – the music recognition App. This has a setting which allows it to constantly listen in to your world for any music it hears. Some advertisers have already experimented with campaigns where offers appear in the phone automatically if a person is in a store where music tracks are playing. So the phone is then using music to locate precisely where the person is, in co-operation with marketers.

Thought-control and bias confirmation by Artificial Intelligence

One of the most worrying aspects of Artificial Intelligence is the ability to influence people's opinions and decisions without them realizing. It's already happening across the globe.

Let's go back to the AI example above of semantic search.

Imagine that a student has been asked to write an essay on reasons why global warming science may not be as accurate as many people think. So he does hundreds of searches using terms like "global warming fraud", or "global warming conspiracy" and so on.

So Google AI in it's search engine will make a note that this student is particularly interested in arguments against global warming, as evidenced not only by search terms, but also which pages he or she spends most time reading.

Two years later, the same student hears about some new research that suggests global warming is happening much faster than previously thought. He enters the search request:

"New research confirming global warming"

Because of the previous history, Google filters the results quite heavily. The top articles are all aggressive critiques of the piece of research by well-known climate change sceptics, most of whom have few scientific credentials, but who have huge social media followings.

Right down the listings are articles written by the scientists themselves, promoting their research, plus articles supportive of it by other scientists.

The student may not be aware at all of a strong editorial bias in what Google is presenting, and may be seduced into assuming that the latest piece of research is discredited and worthless, when the exact opposite may be the truth.

So what is happening – and this is a wide trend across platforms like Facebook as well – is that hundreds of millions of people are having their own biases and prejudices confirmed by distorted news feeds, info sources and research results.

Maybe, you say, opinions of one person on global warming don't matter too much. Well what about a stream of articles and news posts that appears to confirm the very worst and most outrageous racist attitudes of someone who then goes on to assault and kill someone because of the colour of their skin?

You might think this kind of Artificial Intelligence mind-bending is far fetched but it is exactly what we are seeing in radicalization of young people with certain backgrounds, in a journey towards joining extremist groups like ISIS, some of whom will end up committing atrocities or acts of terrorism.

Account blocking and total exclusion from normal society by AI robotics

China has already introduced new web-based scoring systems, controlled by Artificial Intelligence, which control web access for its citizens. Those with highest scores for responsible citizenship are going to be rewarded with fastest access, while those with lowest scores will be almost completely or totally excluded from the online world.

Similar things are happening across other nations, even in those that regard themselves as having the highest ratings for personal freedom of expression and for lack of online censorship or controls.

If we return to the example above of a student who is researching – say – ISIS and radicalization, rather than global warming. There has been a big change recently in government attitudes to information hosting services. In the past, companies like Facebook said that they were just platforms onto which people can freely post content. They are not publishers as such and so do not need to take responsibility for what content is posted, whether it is accurate, harmful, libelous and so on.

But government attitudes have hardened because so much of online content is now morally degrading, repulsive and damaging - not only to adults but in particular to teenagers and young children.

So companies like Twitter, Facebook, LinkedIn, Amazon, eBay and Google are facing the prospect of massive fines in future, if they fail in their duty to take down a wide range of content that governments decide should not be permitted.

Wider than this, such companies are under pressure to act with great speed, and to lock suspect accounts. And presumably to share information with other companies so that a combined action can be taken across all major platforms, to stop the same person from just moving their terrible content around.

So, imagine a student who is researching into radicalization, who views various awful websites as a legitimate and essential learning process. Immediately the person may become a "person of interest" to Google's own AI. The same person made a couple of Facebook posts about their research, one of which was posted by them on their LinkedIn page and also on their Twitter account. They posted a summary on their personal web page – and suddenly the combined AI machines at Google, Facebook, Twitter, LinkedIn made a collective decision. Their Facebook account was blocked, and seconds later their Twitter account was closed and the listing of their blog on Google was also deleted. Their LinkedIn account was also suspended without warning.

Not a single human being has been involved. But you can be sure that a human being will be required to review and possibly lift all these blocks. And how long will that take? Will it even be possible? Since most or all of these services are free, what moral right does the individual to complain about lack of service or exclusion for a few days, weeks or months?

But his or her daily life now depends on these same platforms. The student finds that job hunting is not completely impossible because employers find a page saying "Account Suspended due to Violations" when they type the student's name into LinkedIn. The small business the student has been running selling things on EBay has also been suspended, and Amazon account blocked.

We are very, very close already to such scenarios. And they become more far-reaching every day because of consolidation and mergers of online companies. So for example, Google also owns YouTube. LinkedIn also owns Slideshare. Facebook also owns SnapChat.

Only one small step further and we have a situation where AI alerts from things like the above may be powerful and convincing enough to damage credit ratings, the ability to go on using a credit card, or to withdraw cash from the bank.

Automated journalism by Artificial Intelligence

Part of the AI news-influenced world is the growth of AI journalists. These are software programmes designed to take data sources such as recent government publications, and summarise them into simple news headlines and a few sentences of text.

Once again there are opportunities for inadvertent or deliberate bias, and also for cascade reactions when such AI-produced headlines start triggering massive AI-controlled selling decisions in the markets. And then the world starts to get very complicated.

Drone assassinations, autonomous weapons and AI linked wars

Every Artificial Intelligence trend is related in some way to every other AI or wider digital trend. So the growth of drone technology, and growth of face-matching, together with other data to locate people, is now making possible the growth of autonomous killing machines and AI wars.

A drone is in the air, carrying a very simple weapon – maybe something like a hand grenade with a pin that the drone can remove, or maybe just the blades themselves, made of ultra sharp steep, with the strength to instantly decapitate a man or woman.

The drone has a very high-resolution camera, and is constantly face-matching using AI support from the web. It is also being fed all kinds of other AI data to help it track and confirm human targets, any one of whom it is authorized in advance to kill on sight without any further human involvement.

Do such weapons already exist? The answer is yes, in experimental form. All the elements are already available, and just need to be assembled into one AI autonomous weapon system.

So how would this impact the future of battlefield conflicts? It is easy to envisage swarms of drones, operating freely over large areas, making instant decisions to attack enemy equipment, vehicles or personnel. As part of this, there are obvious risks of error and of war crimes or crimes in general against humanity.

Recent tensions between North Korea and the United States have highlighted many other AI-related risks, where systems may be reporting a suspected nuclear weapon launch etc, that has not actually taken place. How reliable and smart is North Korea's Artificial Intelligence when it comes to reporting that their nation is under attack? What would happen if North Korea's AI made a massive error which triggered an immediate launch of a nuclear missile against the US, South Korea or Japan?

Biodigital human brains connected to Artificial Intelligence

Some say that the ultimate Artificial Intelligence is a globally-empowered super-conscious, self-learning machine which effectively starts to manage life on earth.

But I am much more fascinated by a more immediate opportunity, which is to link AI directly into the human brain – in both directions.

This would enable you or I to have access to additional information or processing power, as a direct extension of our natural ability to think or remember. And it could also provide AI with additional wisdom, from the collective consciousness of millions of people.

It is becoming quite normal to fuse computer chips with human brain tissue or nerves. Over 400,000 human beings already have chips implanted inside their heads. In most cases these are attached to the auditory nerve to enable deaf people to hear - cochlear implants. The rest are implanted directly onto the surface of the brain, or within the brain, and are being used to deliver data,for example to enable a blind person to see – or receive commands, for example so that a paralysed person can move their arms or control a machine.

All these implants are primitive today, and chips touching brain tissue carry a risk of triggering epileptic fits, so we need a really compelling reason to use them today. But what about tomorrow?

Thought detection by Artificial Intelligence

In the meantime, IT companies are investing a lot in headsets of various kinds which can pick up tiny electrical brain waves, using AI to detect common patterns, which can indicate what the person is thinking about.

At the most basic level, this means a player of a computer game can control screen characters or weapons by thinking alone. At a more sophisticated level, such devices may reveal in future what a person is imagining, or thinking about.

We are already using medical brain scanners to study how consumers respond to advertising. Neuro-imaging has led to Neuro-marketing. We can watch in real time how areas of the brain light up in response to particular images, colours, smells and brand straplines.

Will AI or robots destroy jobs? Impact on workplace

In the media everyday we see sensational headlines such as "Hundreds of millions of jobs will be lost to AI and robots", suggesting that humankind will never be able to find enough work for tomorrow's people, because so many jobs will be done by Artificial Intelligence machines.

In the UK, the Bank of England predicted in 2015 that 15 million UK jobs would be lost to robots.

The Truth about Robotics is this

1) Every industrial revolution in human history has benefitted humankind in the same way. Large scale mechanisation, innovation, automation, industrialisation - these things cut costs and so are widely adopted. The biggest cost they reduce is labour. Fewer people, working more efficiently, produce more. 

2) As costs fall, standards of living rise. Society as a whole becomes wealthier.

3) Many traditional jobs start to disappear - usually relatively low grade, manual jobs but more recently, executive jobs are also being hit.

4) At the same time, human ingenuity, innovation and creativity is constantly searching for new ways to make life better, make people happier, make the world a better place to live in.

5) And most of these new initiatives fail because although they are very worthwhile, although many people really want them, we can't really afford them - at least very few people can.

6) But as costs of many things fall the situation changes, because of automation, people and governments begin to find they have enough money to spend on these new things. They also find there are plenty of workers available to provide these goods and services.

7) New markets begin to form, and unemployment stabilises.

Now in practice, all of these 7 factors are operating at the same time to different degrees in each nation, and tend to even each other out.

Shift from automated labour to services

The biggest shift over the longer term is always a migration of labour from production (easily automated) to services (where human touch is often more important).

That has certainly been the case in nations like the UK which have very small numbers of workers in manufacturing, producing more goods than in their entire history, yet have very low unemployment because of the transition of workers into service jobs which never existed before.

Major job losses usually driven by economic cycles not automation

Yes we have seen major rises in unemployment recently in many nations, but almost always linked to economic cycles rather than industrialised.

The PRODUCTIVITY paradox

For years, many governments like that in the UK have been complaining that productivity in the workplace has not been growing fast enough, or hardly at all. They base this on a wide range of economic data.

These same people also declare that they are worried about robots and AI destroying hundreds of millions of jobs. But how can both these concerns be reconciled?

Either we are seeing very little impact from the entire digital revolution including the Internet of Things, Big Data, the Cloud, Robotics, automation and Artificial Intelligence – or we are seeing an impact, but one which current systems are not picking up.

It is impossible to believe either extreme. 

Clearly the digital revolution saves us all time, money and makes society more efficient. Just take a simple example of being able to inform colleagues that you are going to be late for a meeting, or being able to send a text instead of posting a letter, or being able to copy and paste sections of one long document into another.

And clearly the combined impact of all this on unemployment rates has so far been very small indeed. The fact is that in the UK, despite 10 years of economic challenges following 2008/2009, and Brexit uncertainties, more people were in jobs than ever in history and levels of unemployment were lower than for 30 years. Hardly a picture of an AI apocalypse in the workplace.

The ROBOT Paradox

Here is another AI or robotics paradox: many people who say they fear mass wipeout of millions of jobs, are also the same people who say they fear that society will never be able to afford things like looking after ageing populations properly. 

How so?

Automation will help created the wealth we need to be able to pay all those additional carers. Automation means more incomes, more profits, more taxes and more money for governments to spend. It also means more disposable incomes for ordinary people.

We see this effect even at times of low salary growth or inflation. Take the last decade - a period when incomes in many nations have been relatively static, or have even fallen in real terms if you take inflation into account. 

During the same period, many of the things that people like to own or do have become much more affordable, because of automation / industrialisation / scale. 

Examples include buying a basic smartphone, cost of a fast web connection or of long distance calls, cost of taking a budget flight or a (Uber) taxi, or of taking a city break (AirBnB), or of buying basic groceries (Lidl and other bid discount stores).

Big impact from automation

So then: watch out for big impact of automation in every sector, every industry, every office and factory - but take a look at innovations in new kinds of services even day - leisure industry for example, with huge growth in coffee bars and restaurants, and of every kind of family entertainment including theme parks and leisure centres.

Machine consciousness and super-human awareness

So, in summary then, however you try to define Artificial Intelligence, the truth is that machines are becoming rapidly more intelligent, in a process that is advancing many times faster than the growth of computing power or the growth of the web.

AI poses some of the most profound ethical and philosophical questions that humankind has ever faced, and will deeply transform how society lives, works, thinks and feels.

At the same time, many of the wildest fears that people may have about AI are just that – wild, and probably further off in reality than they imagine.

In the meantime, we should be watchful, responsible, debate the issues, harness all the power for good in AI, and take care to manage the bad.

* Patrick Dixon is a keynote speaker on Artificial Intelligence, AI impact on industry, manufacturing, health care and wider society.

< Prev		Next >