It’s been some years since I seriously attempted to study software development in any degree of detail, however, I still remember vividly thinking “humans will never make a viable AI (artificial intelligence)”.
Having a basic understanding of how software worked, how machines crunched problems, along with a general understanding of our own lack of understanding regarding the fundamentals of consciousness. I was sure that AI would simply be a problem that we wouldn’t have to face in our life time, if any.
I could not have been more wrong.
And it turns out that I wasn’t alone.
Sergey Brin, co-founder of Google himself has come out remarking how he too is surprised by the sheer rate of development in recent history.
We have to make it clear, with the advent of modern AI, we’re looking down the barrel of an economic change unlike the world has seen since the very first industrial revolution. AI will change everything. AI will shape our future economy in ways that we can’t even begin to imagine.
But to understand how, we first need to take a journey back in time to understand how the previous industrial revolution(s) have affected our society.
Pre 1750’s, the notion of economic growth was an unheard of concept. This was a period of time described by economists as a period of a zero-sum game. Society was divided into two distinct classes, wealth was tied to political power and life for the average man, save for a tiny commodity trading middle class, was by all modern standards, poor.
The income earning potential for the proletariat was directly linked and limited to their labour capacity.
That is until roughly 1750 where the Industrial Revolution rocketed England into a new direction. Through the power of steam, over the next 200+ years, biological muscles were replaced with machine muscles. Muscles that wouldn’t tire, wouldn’t fatigue, that didn’t produce the same amount of waste and were in many ways greener on an output for input measurement.
Eventually, these metal muscles moved from the vast factory floors in the seventeen and eighteen hundreds, to revolutionising transport and even home care even down to washing our dishes and cooking our food. Mechanical maids, butlers, servants that were once were limited to the affluent became the norm for the lower class and with this new ‘slave’, the demand and value for much of the labour that previously had to be done by biological hands, plummeted while at the same time the value of the intellectual capacity increased as these mechanical muscles lacked the intelligence of a two-year-old.
The marginal productivity of an individual worker started to climb as workers were better able to spend their time, meaning that the world experienced real, significant economic growth, growth not tied to population growth. Growth that wasn’t tied to any banking trickery or money printing scams. Growth that was the product of simply better uses of resources. Growth that was a product of increasing productivity, a growth that came from the enslavement of metal to serve our desires.
The industrial revolution marked a fundamental change in society, firstly paving the way for a huge population increase, a concentration of population towards metropolitan areas, and eventually lead to a whole new class of people being born, that was, the western middle class. A class of people who relied not so much on their sweat to earn a dollar but rather their mental sweat, their ability to control the machines that raised their marginal productivity many folds. Our middle class came reliant on the fact that there was simply some tasks that machines could not do, and the demand for those jobs swelled.
Overtime, due to this ‘outsourcing’ of labour, a new ‘knowledge’ based economy was born and has continued to grow year on year especially with the advent of communication and computer age. Now, one of the greatest determinants of potential income earning potential is your IQ, and thanks to computers and virtual products, we have seen the advent of a new class, the class whose income is only limited by their knowledge, determination, and imagination. Cold hard industrialised, computerised capitalism gave birth to a true meritocracy.
That is, until now.
Just as the mechanical muscles replaced the far inferior, far less efficient, biological muscles, so too will AI replace biological brains. The jobs that we’ve for a long time considered outside the realms of machine capability are becoming ever more increasingly within their grasp.
The way in which this new age of machine capacity is coming about is largely through a technology known as machine learning. Without going too much into the technicality of how this works, machine learning is the process where instead of giving the machines the answers to every problem, we instead get the machine to figure out what the right answer is based on past data sets and with the incredible growth in modern computing power, we now have the theory and the hardware to build ever increasingly powerful AIs.
The results of this is that instead of programmers having to feed software thousands of lines of code specifying what is and what isn’t a bird, the computer that is implementing this technology would be able to learn what a bird is. The computer can ‘learn’ what is a bird by identifying what the common characteristics are much in the same way to how humans learn. AI gives computers the ability to make sense of the world from the binary data that is fed to it.
This ability to process a given input and through ‘experience’ be able to figure out the best possible output, with data sets far larger and theoretically more complex than what humans can handle, is literally groundbreaking for a number of reasons.
What can we do with it?
To some extent, you’re most likely already reaping the benefit of this technology. Every time you ask Siri for the weather or to set a reminder, you’re benefiting from the Siri application being fed huge data sets. Every time your camera phone uses facial recognition or Netflix suggests a movie to you, you are experiencing the benefits of AI, even if it is in its infantile stages.
But, like all technology in its infancy, the applications doesn’t and won’t stop there. Instead, the jobs and applications that these machines will master will start with the ‘simpler’ tasks where data is easily collectible on massive scales and will move into tasks where the data collection is less structured.
With the use of complex scene recognition, driverless cars have come leaps and bounds to such an extent that viable consumer products already exist and every major car company is currently working on commercially viable self-driving cars to be on the shelves within the next 1-3 years.
Digital translators. Microsoft has already developed an application that will, in real time, translate between two or more individuals using different languages, where possible using each speaker’s morphology and accents.
Humanoid robots, they aren’t that great yet… but watch this space. Massive data collection agencies like Facebook, Instagram, Google etc, mean that AI’s have all the big data they need to start analysing, studying and learning about human interactions, combined with a year on year betterment of AI language development, realistic hominoid robots are not that far away. Below is a video that discusses Facebook’s work in this area. With unlimited access to over a billion user videos, photos, and posts, worldwide, I can guarantee that they won’t limit this technology to describing the content of images to the visually impaired.
Creative machines, yep you heard right! We now have AI’s that are capable of painting pictures, creating animation, writing music and even generating adverts of which Coca-Cola is already heavily experimenting with.
Farming. One company in the USA is already utilising AI to improve crop yields for the tomato farms by being able to automatically assess plant health and potential issues.
Medicine. The list of potential applications is far too long to list here but the fundamentals of medicine aren’t that complex, that is to say, doctors are typically pattern recognition machines. Through study and experience, doctors take in huge data samples and then create a prognosis based on the most likely causes. With AI, the potential input datasets can be much greater than what a human can process ultimately leading to a much more accurate diagnosis.
Journalism. Unfortunately even the more “creative” industries that we have for a long time considered off limits to computer automation, are now within a computer’s grasp… and let’s be honest, could these computers really be worse than many of our current journalists?
From Law, Finance, Factories, Farming to the creative industries. There is no conceivable field of human work that isn’t off limits to being radically changed by the AI revolution. Just as mechanical muscles have changed our economy and our society from the ground up, so too will mechanical brains change our world in ways in which we cannot even begin to imagine.
“At this point, with the technology that we know about today, we are probably talking more in the data entry areas. We still do a lot of very manual processes in the bank and, as I said, machine learning is pretty remarkable today. A machine can read a pay slip, identify what it is, take the data and stick it into a model for us. I don’t know. All we are saying is: I think we’d be naive to say that technology is not going to fundamentally change the nature of work right across the economy, and that will absolutely include banks.” – Shayne Elliot Chief Executive Officer, Australia and New Zealand Banking Group Limited answer questions in the Standing Committee on Economics (11/10/2017), Review of Australia’s four major banks.
However, speculation, of course, is fun and crucial for being able to navigate this brave new world.
What does it mean for human workers?
This is the point where we move from what is currently in the world of development to what the likely affects will be in our day to day lives. These predictions must be broken down into the short-term and the long-term effects.
By short term, I don’t mean to specify a period of time in years, but rather use the term to describe a period in which this technology is developing and being implemented into the real world applications with the long-term being the next plateau of implementation and development.
We have to make two things clear at this stage. Firstly, these machines will, bit by bit, take over currently available jobs. By way of example, the transport industry employs over 300million people worldwide and while not all of those jobs are under immediate threat, the rate in which humans will be made redundant will be at a pace much greater than the rate in which new jobs are created.
Which leads us to our second observation, that is, not all development leads to more jobs and more economic activity in total. The smartphone industry can provide a perfect example of destructive innovation. While smartphones have enriched our lives, empowered us with a whole new level of communication possibilities, they have also lead to the destruction of many other industries including pocket cameras, the personal PDA, the mobile phone, the Gameboy and to a large extent even the laptop market. And while the end product is better for most consumers, the total value of the smartphone industry would be less than the total economic value of all those other industries that it consumed. The reality is that AI has a much greater general applicability than the smartphone (as useful as it is), meaning that AI will consume more jobs, more economic activity and hence take more tax dollars from government coffers than what it will return.
In the short run, the economic effects will be a decrease in the value of human labour across the board due to automation and AI development. This will start to create structural unemployment and grow our government deficits as more and more workers move from being taxpayers to welfare recipients. However, at the same time, we should start to see a real price reduction in the cost of goods and services as our production lines become more efficient and as the real wages decrease. Without government intervention, this will result in an economic deflationary period and a recession.
Governments at this stage will have to react. There will most likely be great social pressure to ban certain AI, automation etc, but due to large corporate influence and the threat of ‘being left behind’ by overseas competitors, governments will most likely opt to modify the welfare and taxation system. Ideas like the Universal Basic Income will most likely be adopted in certain wealthy countries from one degree or another.
In the meantime, while labour becomes relatively less valuable due to a substitution effect, capital will become enormously more valuable. During this time, we will most likely see a significant reduction in the size and wealth of the middle class.
Our economy relies on consumption. Fait Currency doesn’t hold intrinsic value. Due to the deflationary effects of the structural unemployment caused by this machine revolution, combined with the advances in real productivity (possibly the greatest increase in productive capacity for a single unit of labour since the first industrial revolution), ultra low-end jobs that were previously considered too expensive for humans undertake throughout the 20th century, will start to become viable with greater wages, as time and money are released from other parts of the economy.
There will almost always be a tradeoff between the cost of employing a machine to perform a certain task and employing a human to undertake the same task, and due to the nature of certain jobs, there will almost always be a point where humans will be preferred to fulfill that a certain task rather than machines.
For a long time, many of our day to day activities and tasks fall well beyond the spectrum of what is economically viable to outsource to a machine, even though in some instances the technology already exists to outsource. As machines become more and more capable in ways that we can only begin to imagine, the spectrum of what falls into the realm of what’s cheaper, employing machines to perform a task as opposed to employing a human to carry out the same task falls into the realm of sheer speculation. What is certain, those wielding the machines, those with the capital, have the capacity to become relatively more powerful and wealthier relative to the already shrinking western middle classes. The AI revolution will likely lead to greater wealth disparity and rob individuals of their economic self-determinism. That’s a pretty dark prediction but one which I can’t really foresee playing out any other way. Either that or, the degree to which these machines will develop is way overestimated and the job losses will be limited to very basic tasks.
I’m betting and preparing for the former.