Three Things Techno-Religionists Get Wrong
Are humans any less religious today than they were 300 years ago?
It depends, of course, what you mean by religious. But it’s worth bearing in mind that 300 years is, in evolutionary terms, the blink of an eye. Our brains have not changed in that time and therefore we can be sure of at least one thing: our capacity for religious belief has not altered. So what are we non-believers doing with that spare capacity, now that it’s not busy maintaining our faith in the supernatural?
One possibility is that we’ve simply directed our religious habits of mind towards a different focus: technology. This, I will argue, is a problem because it leads us to misunderstand the impact technology has on our lives in three important ways.
Myth 1: The god-like autonomy of machines
One important aspect of religious belief is the ascription of autonomy and power to forces perceived to be outside human control. In this respect, our contemporary view of technology is religious in the extreme.
Karl Marx spotted this 150 years ago. In Das Kapital, he wrote that ‘[in] the misty realm of religion … the products of the human brain appear as autonomous figures endowed with a life of their own… So it is in the world of commodities with the products of men’s hands.’ Marx called this tendency fetishism and it’s not hard to spot examples of the fetishisation of technology today — especially when it comes to discussions about Artificial Intelligence (AI).
Hollywood has been fetishising AI for decades — think HAL 9000 in 2001: A Space Odyssey or Skynet in the Terminator films. Elon Musk has likened AI to ‘summoning the demon.’ Kevin Kelly — one of the foremost prophets of the new techno-religion — has laid out the fundamental tenets of the faith in his books What Technology Wants and The Inevitable: Understanding The 12 Technological Forces That Will Shape Our Future.
Fetishism is rife amongst both techno-utopians and techno-pessimists: whether you think AI is going to deliver us safe to the shores of abundance or into the jaws of joblessness, you’re guilty of it.
There are good reasons to be wary of fetishism: as Oliver Morton of The Economist writes in Megatech: Technology in 2050, ‘technology can never be relied on to solve problems in the absence of social action; one of the dangers of fetishising technology as an actor in its own right is that it obscures this point.’
But this doesn’t mean we should ignore the role of technology in modern society altogether — merely that we should strip it of its divine characteristics. AI will have — in fact, is already having — a profound impact on our economy and society, but the relationship goes both ways. We’re obsessed with the question of what AI will do to capitalism; we should also ask what capitalism is doing to AI.
A few years ago, Jeff Hammerbacher, an early Facebook employee who became disenchanted with the world of big tech, lamented the fact that ‘the best minds of my generation are thinking about how to make people click ads.’ Given the way economic power is distributed today, we risk the same being true of the best artificial minds of our generation.
Myth 2: The otherness of the future
Another common feature of religions is their fascination with what lies beyond the knowable horizon. We want to know what comes after death. For all that they may be grounded in ancient scriptures, all the major faith traditions keep the eyes of their true believers firmly fixed on the future and the rewards that they can hope to attain if they live a holy life.
Here, again, there are parallels with the way many people view technology. The singularity (the moment when a machine can outperform humans on any intellectual task) is to techno-religionists what the Day of Judgement is to Christians.
Whether or not you believe, with Ray Kurzweil, that the singularity is near, this mode of thought has infected most of the debate around technology and its impacts. We ask, incessantly, what will happen, rarely stopping to look at what is happening. The discontinuities ahead are so great, we are told, that the past is no guide to the future.
This, again, is a problematic view because it blinds us to the fact that our economy is not so much on the brink of an AI-powered seismic shift: rather, we’re a good 25–30 years into the revolution. As Pedro Domingos writes in The Master Algorithm:
‘[Machine learning’s] first big hit was in finance, predicting stock ups and downs, starting in the late 1980s. The next wave was mining corporate databases, which by the mid-1990s were starting to grow quite large, and in areas like direct marketing, customer relationship management, credit scoring, and fraud detection. Then came the web and e-commerce, where automated personalisation became de rigueur. When the dot-com bust temporarily curtailed that, the use of learning for web search and ad placement took off. For better or worse, the 9/11 attacks put machine learning in the front line of the war on terror. Web 2.0 brought in a swath of new applications, from mining social networks to figuring out what bloggers are saying about your products… Today, there seems to be hardly an area of human endeavour untouched by machine learning.’
So what has been the economic impact of this revolution in its first 30 years? The Economist’s Adrian Wooldridge, writing in Megatech, sums it up in a single sentence: ‘almost all the productivity gains of the past 30 years have been gobbled up by the richest 1%.’
Now, we should, of course, avoid the fetishisation trap. Technology did not make this massively unequal distribution of the gains of growth inevitable. A different set of political choices about how to regulate the economy could have led to a very different outcome. But technology has certainly been a key enabler of rampant inequality — not least because automation has already largely destroyed the link between job creation and wealth creation that underpinned the decades of shared prosperity that followed World War Two.
Consider the following numbers, from an Economist special report published in 2016:
‘In 1990 the top three carmakers in Detroit between them had nominal revenues of $250 billion, a market capitalisation of $36 billion and 1.2m employees. In 2014 the top three companies in Silicon Valley had revenues of $247 billion and a market capitalisation of over $1 trillion but just 137,000 employees.’
In other words, today’s big tech firms are generating equivalent revenues and a market capitalisation almost thirty times higher than the automotive giants of a quarter century ago, with just a tenth of the workforce. No wonder Silicon Valley bigwigs are starting to worry about becoming the new bankers: loathed for their role in creating a seemingly unbridgeable gulf between the top 1% and everyone else.
CEOs earning several hundred times more than their average employees — in 2015, the ratio for S&P 500 companies was a whopping 340 — has provoked justified anger. But this isn’t the real story. As Nicholas Bloom noted in a Harvard Business Review cover story last year, it’s inequality between firms — not inequality within firms — that is responsible for most of the increase in income inequality in the US over the last 30 years. Similarly, Bank of England Chief Economist Andy Haldane has shown that the top 5% of firms in the UK have seen strong productivity growth in recent years, whilst the productivity of the other 95% has flatlined.
In short, the big winners of the past 30 years have been a handful of big tech firms in the US and China: Apple, Alphabet, Amazon, Facebook, Microsoft, Tencent, Alibaba, Baidu. There is every reason to believe that, as AI in particular becomes even more pervasive in the next 30 years, the economic power of these companies will become even more firmly entrenched.
After all, they have amassed in unprecedented quantities the two things most crucial for success in the AI era: money and data. As Mariana Mazzucato notes in her new book, The Value of Everything, ‘just five US companies (Google, Microsoft, Amazon, Facebook and IBM) own most of the world’s data, with China’s Baidu being the only foreign company coming close.’ IDC, a research firm, predicts that by 2020 the market for machine learning will reach $40 billion and that 60% of AI applications will run on the platforms of just 4 companies: Amazon, Google, IBM and Microsoft.
Myth 3: Abundance
Which brings us to the third and final myth promulgated by the techno-religionists: that we are headed towards a future characterised by abundance. The land of milk and honey is to be our reward for holding strong in our faith in the machines.
But the important question, as should be clear by now, is: how will that abundance — assuming it materialises — be distributed? Here, the story is somewhat mixed.
Lest you’re thinking by this point that I’m a pessimist through and through, let me start by saying this: when it comes to energy, I think there are good reasons to believe we are headed towards a world in which clean energy is abundant and distributed. And that is a real cause for celebration.
In other areas, I am less optimistic. Data is already one of the key sources of value in our economy — and its importance is set to grow in the years ahead. It will undoubtedly be abundant, but unless we find the political will to democratise its ownership, most of us won’t reap the benefits of this abundance.
As Calum Chace argues in his book, The Economic Singularity:
‘Scarcity hasn’t disappeared: it has changed, and become more dangerous. The new scarcity is the privileged access to an accelerating flow of powerful new enhancement technologies. The danger is that the elite which enjoys this privileged access will rapidly become a separate species — a dominant species… The new scarce resource — the privileged access to the cascade of new technologies — is even more valuable than any scarce resource that we value today.’
Chace also makes the point that technological progress is at least as likely to undermine future growth as it is to unleash a new golden age of prosperity. This may seem counter-intuitive given that we know technological advances are the key factor in productivity growth, which in turn is critical for economic growth. But productivity is only half the story: without growing demand, productivity gains simply fuel a negative feedback loop of rising inequality, falling demand and stagnating growth.
Already, growing inequality has led to lower aggregate demand and therefore slower growth across the Western world. The relationship between inequality and demand is really very simple. As Joseph Stiglitz writes in The Price of Inequality, ‘moving money from the bottom to the top lowers consumption because higher-income individuals consume a smaller proportion of their income than do lower-income individuals.’
The full effect of rising inequality on growth has been masked by a series of asset bubbles (tech in the 1990s; the US housing market in the 2000s), central banks’ quantitative easing schemes since the 2008 crash, and a decades-long build-up of both private and public sector debt, which today stands at $164 trillion — or 225% of global GDP — according to the International Monetary Fund.
So we’ve backed ourselves into a corner: will AI help us get out of it?
Looking at a range of AI techniques and their applications across 19 different industries, McKinsey estimates that they have the potential to create between $3.5 trillion and $5.8 trillion in value annually. Add in analytics techniques that do not rely on AI and they predict the overall economic impact could be as much as $9.5 trillion to $15.4 trillion a year.
The argument that AI will create trillions of dollars of value is the orthodoxy in Silicon Valley. Microsoft CEO Satya Nadella may worry out loud about how the surplus created by breakthroughs in AI will be distributed, but he seems pretty certain that a surplus there will be.
But this view ignores the impact AI will have — indeed, already is having — on the demand side. Here’s what Chace has to say on the subject:
‘If machine intelligence renders more and more people unemployable, then other things being equal, the purchasing power exercised by those people will dry up. Their productive output will not be lost — it will just be provided by machines instead of humans. As demand falls but supply remains stable, prices will fall. At first, the falling prices may not be too much of a problem for firms and their owners, as the machines will be more efficient than the humans they replaced… But as more and more people become unemployed, the consequent fall in demand will overtake the price reductions enabled by greater efficiency. Economic contraction is pretty much inevitable.’
Universal Basic Income (UBI) is the de rigueur solution to this problem of rising inequality and falling demand. But the announcement last month that the Finnish Government would not be extending its flagship UBI trial on grounds of cost has dealt the idea a heavy blow.
In short, the sunlit uplands of abundance look rather a long way away — and whether we ever make it there actually has relatively little to do with technology. Unless we solve the social and political challenge of how to create a more distributive economy, all the AI in the world will do us about as much good as an umbrella in a war zone.
