Monday , July 22 2019
Home / Energy / The real threat from AI

The real threat from AI

Image: David Blackwell

Killer robots really are coming for your children… expect to be plunged into a new Dark Age!

Am I thinking of some kind of dystopian sci-fi future in which an army of Terminator-type robots arrives to exterminate humanity?  Or maybe I am imagining a near future in which Jeff Bezos, Bill Gates and Mark Zuckerberg have us all hard-wired with a brain-chip that connects us to some Borg-like singularity?  Actually, no, I am thinking of something far more mundane.

While techno-utopian fantasists imagine some kind of hi-tech Star Trek future, those of us who have taken the time to look at the precarious state of our electricity grids and the fossil fuels that power them can see an altogether different future. 

Our electricity infrastructure already ranks among the most complex machines ever built.  And remarkably, until recently it was pretty damned reliable… far more so than the system when I was a child, when intermittent power outages were a fact of life.  Despite their complexity and potential fragility, politicians and economists have determined that we have to dispense with the majority of the generation infrastructure and replace it with tens of thousands of intermittent non-renewable renewable energy-harvesting devices whose penetration of the system is already creating problems for grid engineers charged with maintaining a continuous balanced load; not to mention the additional cost to already struggling businesses and households.  Not content with this, the politicians and economists have decided that the grid will have to provide all of the additional electricity needed to swap internal combustion engine cars for new electric vehicles; to electrify a large additional part of the rail network; and to power a large part of current fossil fuelled industrial processes.  It would be an understatement to suggest that all of this comes with some engineering challenges, and will likely come to grief if supplies and or/affordability of gas imports becomes an issue.

None of this is even considered by the proponents of the “4th industrial revolution,” who imagine that it will be easy to maintain a civilisation built upon fossil fuels without continuing to use fossil fuels.  And yet people in the technosphere should be more aware of the potential threats than we mere mortals.  After all, data centres need a constant supply of power to prevent them overheating and burning out the hardware.  That hardware, in turn, is manufactured in plants that require continuous power to create enclosed positive pressure units to prevent contaminants like dust entering the facility.  To blithely assume that a constant power supply can be maintained when we cease using fossil fuels – even as our appetite for more power is growing exponentially – smacks of gross negligence.

The bigger immediate issue, however, concerns the economics of technology.  Much of the imagined future of colonies on Mars, hyperloops, self-driving electric cars and a new knowledge economy depends ultimately upon economic conditions almost entirely opposite to the conditions we are currently facing.  Private sector automation requires that we have labour shortages coupled to a fast growing consumer economy.  This provides corporations with an incentive to invest in the technological upgrades required to automate as many roles as possible. 

On paper – but not in the real world – we have the first condition; full-employment.  Unfortunately, this has only been achieved by stretching the definition of “employment” to the limit of credibility.  In the UK, for example, someone sat at home all day doing nothing can be counted as “employed” if they have a zero-hours contract with a company that has no actual work for them to do.

The second condition – a fast-growing economy – cannot even be created on paper.  Since 2008, the economy has been anaemic; generating levels of GDP growth that would have been considered an economic crisis in the years before the crash.  Not only have we witnessed a retail apocalypse, but all the indicators suggest that the crisis is accelerating and cascading into new sectors of the economy like car sales and housing.  In these conditions, corporations are better off buying back their own shares than investing their profits in technologies that are unlikely to improve their bottom lines.

For the moment, the tech sector – like the fracking industry – is enjoying Wall Street largesse simply because the low interest rate environment encourages investors to take risks.  The problem is that this tends not to be investment in the roll out of proven technologies to real businesses that make real things.  Rather the investment has gone into things like solar roadways, self-driving cars and hyperloops that will – like the fracking industry – prove to be worthless once interest rates begin to rise.

Rather than the imagined economy of the hi-tech 4th industrial revolution, it is the low-paid/low-tech public services sector that has seen the biggest shift to automation.  As James Howard Kunstler has pointed out, we now have the most advanced communications system in the known universe but it is harder than ever to speak to an actual person.  Banks, insurance companies, utilities and transport companies have all installed automated systems for handling customer inquiries and customer complaints; allowing these companies to cut costs by laying off already low-paid workers… workers who also become consumers when they are off duty… the kind of consumers who can no longer afford to take out loans or buy insurance policies, utility services or travel tickets because someone fired them.

The big growth area for AI and bots, however, is the public sector where government spending cuts are forcing service managers to use automation to cut their wage bill.  Robotic process automation (RPA) turns out to be well placed to deal with the levels of bureaucracy and form-filling that only governments seem able to devise, as a PR article in the Guardian notes:

“RPA replicates the same tasks carried out by a human worker, but executes them far more quickly and accurately. The bot acts as a substitute for the human worker. It has a password to log on to a system, then goes through the same tasks a human worker carries out, such as opening a database and entering data from a spreadsheet. The bot can be programmed with rules to carry out a task, or it can learn the processes by replicating the keystrokes of a human worker and the screen positions they use to complete the work.”

Across the UK, councils, Health boards and government agencies have been automating this kind of routine work as a means of cutting labour costs.  In the process they have replicated all of the drawbacks in the telecommunications system… it is hi-tech but you cannot actually speak to someone.  So long as your enquiry fits within the parameters programmed into the system, you’ll be fine.  But heaven help you if you need something unforeseen.

Automation is also likely to creep out into other areas of the public sector.  Last year Henry Bodkin at the Telegraph reported on the use of AI in education:

“Robots will begin replacing teachers in the classroom within the next ten years as part of a revolution in one-to-one learning, a leading educationalist has predicted…

“The new era of automated teaching promises an end to grouping children by year, as the personalised nature of the robots will enable pupils to learn new material at their own pace, rather than as part of a class.”

Nor is teaching the only profession that can be at least partially replaced with AI.  In medicine, routine tasks like prescribing medicines and repetitive tasks like analysing test samples are better performed by AI than by a human.  A large part of the work of lawyers involves detailed searches of legislation, previous cases and case law; all of which is better done by AI.

In many areas, the damage has already been done.  In the event of a bank computer crash, for example, there is simply not enough hard cash in the system for people to buy the goods and services they need.  In any case, many retail outlets have already installed computerised tills linked to their bank so that they couldn’t accept cash during an outage if they wanted to.  If we are fortunate, the next round of automation will evaporate along with the value of most of the tech sector when the next financial crisis hits.  Because if ever there was a system built on sand, this is it.

All of us, I suspect, have experienced the frustration involved in losing an internet connection, being unable to access a website or having a computer crash.  Imagine what it will be like in future when such outages become far more common even as the system is far more vulnerable. 

Fast forward to the 2020s.  Britain experiences regular power cuts due to a lack of generating capacity at home and a lack of imports from its European neighbours.  The cost of maintaining the automated systems installed across the economy in the first two decades of the century has spiralled upward due to the high rates of hardware failures resulting from unpredictable power cuts and surges.  To add to our woes, replacement hardware is in short supply and increasingly expensive because of the global spike in demand as grids around the world become less reliable.

Employment collapses because workers and firms are no longer able to timetable the working day because they cannot know in advance if the power supply will be available.  This leads to a string of labour disputes as corporations refuse to pay workers for the time when the power is off, while unions insist that workers should be paid for being at work.

Services that we currently take for granted like telecommunications, banking, utilities and shopping cease being able to function on a just-in-time basis due to power outages and computer failures.  There is, however, no human backup system.  Indeed, in many cases the service has no physical presence but operates entirely across an increasingly unstable internet.

Worse still; in many sectors of the economy nobody knows how things used to be done in the days before computers and AI.  Armchair collapsetarians who theorised that they would be able to revert to horse drawn transport or to survive on tinned beans discover that there aren’t any work horses available and beans cause death (or at least social exclusion) by flatulence.

Key critical infrastructure – including, ironically, the failing electricity grid – was only able to function at 2019 levels of complexity because of the use of computers and AI that are now failing across the system.  The result is a cascading collapse that erodes our civilisation’s life support systems to the point that collapse becomes inevitable.

There is no need for an army of Terminators to show up; still less a Borg-like invasion from space.  We can use AI to destroy our highly complex global civilisation through the simple schoolboy error of installing the tech before we have figured out how we are going to supply it with power.  In the end, this may not matter since the gathering storms of environmental destruction, energy and resource depletion and economic collapse will most likely crash western civilisation anyway.  But I can’t help thinking that the more human systems we can save from the AI, the gentler our descent into the abyss is going to be.

As you made it to the end…

you might consider supporting The Consciousness of Sheep.  There are four ways in which you could help me continue my work.  First – and easiest by far – please share and like this article on social media.  Second follow my page on FacebookThird, sign up for my monthly e-mail digest to ensure you do not miss my posts, and to stay up to date with news about Energy, Environment and Economy more broadly.  Fourth, if you enjoy reading my work and feel able, please leave a tip.

Check Also

Power Station by Martin Kelly

Reductio ad absurdum

One of the advantages of being a rocky island in the northeast Atlantic, right underneath …