On 11th and 12th June, I’ll be attending CogX 2018, Europe’s premier conference on artificial intelligence. One of the speeches I am most looking forward to is Joanna Bryson’s, (Associate Professor, University of Bath) presentation on “Ethical and Responsible AI – What Do We Need to Think About?” at 10:50am on 11th. Ahead of this talk, I have been exploring the issues around ethics in AI, anticipating Bryson’s deeper enlightenment on the topic on the big day.
The automated era is here. Algorithms are already running a large number of everyday aspects of our lives, often behind the scenes where we do not even notice them. This stage, in which we are right now, is only the tip of the iceberg. More and deeper developments are coming, and the ramifications of many of these have the potential to alter the face of modern life forever.
Many experts are therefore calling for serious consideration of ethics in artificial intelligence. This is just a broad term; there are multiple facets to the issue of ethics as they affect different areas of life. In this post, I’ll be covering the concerns of automation and the prospect of a post-work society, with other issues addressed in a series of articles to follow.
The Post-Work Utopia – or Dystopia?
I recently wrote a piece in which I argued that job losses will not be a problem in the age of automation if we collaborate with robots: “The vast majority of us will not lose our jobs and livelihood. Instead, machines will come and work beside us, making our jobs easier and – crucially – more fulfilling.”
But this is only one side of the debate, and – indeed – one side of a many-faced die. Yes, for a few years, collaboration with machines will be useful in the workplace, and will fulfil its promise to take mundane tasks off our hands so we can concentrate on stuff that lets us use our brains. But consider the ever-improving nature of machine learning systems.
In his excellent ebook, The Economic Singularity, Calum Chace postulates the plausible scenario should Moore’s Law continue at its current rate:
“In ten years’ time, our machines will be 128 times smarter than they are today. In two decades, they will be a remarkable 8,000 times smarter than they are today, and in three decades, an astonishing one million times smarter.”
Bearing in mind that many are calling time on Moore’s Law, and that these figures may represent the higher end of the spectrum, we can still conceivably agree that an upward trajectory of machine intelligence is to be expected. Development is hastening, not slowing, and no one is going to stop developing AI now the genie is out of the bottle. In short, the machines are going to get smarter. It may only be able to do the boring bits of your job now, but there is every reason to expect that within a decade (two at most) it will be able to do your entire job as well as you can, and (in Chace’s words) “it will soon do it much better, faster and cheaper than you can”.
This may sound speculative, but it is nonetheless something we ought to prepare for. Saying that it will be okay and sweeping the issue under the carpet won’t help. Whilst, as I noted in my article, the headlines are jam-packed with fear-inducing headlines about job-stealing robots, so much so that I tend to glaze over them or make fun of them more often than not, I am not naive enough as to believe there is no problem here.
The solution that companies, journalists and leaders tend to put forward is to retrain displaced workers. In the short term, we must raise a few objections to this: who will pay for this retraining? What about older workers or those who, for one reason or another, cannot retrain? Who will take care of the financial obligations of these workers in the interim? And will there be enough roles for them to retrain for in the first place? Long term, an answer to this latter question becomes absolutely key.
One of the things people always bring up when talking about the potential unemployment crisis is that new jobs have been created and people have continued to have work in every industrial revolution so far. That every time there’s been a major development like AI, people have panicked about job losses, only for it to turn out not to have been a problem.
On the contrary, the first industrial revolution saw massive upheaval as people’s livelihoods were pulled out from underneath them. Yes, there was a recovery, but it is completely inaccurate to say that this was immediate. There was real, significant suffering and devastating poverty caused by the whole thing. In a December 2017 report by McKinsey, Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation, they state:
“During the Industrial Revolution in England, average real wages stagnated for decades, even as productivity rose. Eventually, wage growth caught up to and then surpassed productivity growth. But the transition period was difficult for individual workers, and eased only after substantial policy reforms.”
Even if we concede that job markets recovered and that productivity and real wages rose at every new significant technological milestone, we cannot rationally consider history to be a reliable gauge for what will happen in the future. “People sometimes say,” muses Chace. “That because fears about unemployment have been raised every time a new wave of automation appears, the people raising the fears are like the boy who cried wolf in the children’s story. They forget that in the story, the wolf did eventually arrive”.People sometimes say that because fears about unemployment have been raised every time a new wave of automation appears, those raising the fears are like the boy who cried wolf. They forget that in the story, the wolf eventually arrivedClick To Tweet
The thing is, with this topic, that there is no conclusive answer yet. From one article I read to the next, my viewpoint veers from one end of the spectrum to the other. The truth of the matter is that we will not know how much effect the introduction of machine learning systems into the workplace will actually have on human employment levels until it’s fully bedded in. And by then, there will be no going back.
This is why the issue of jobs and automation is a matter of ethics. Though we seem to be racing ahead with the introduction of exciting new technologies into the workplace, we are not putting in place the foundations that will uphold the new status quo that is set to develop as the path to automation winds inexorably on. There is much talk of a universal basic income, which sounds ideal in theory, but lacks practicality and foresight. If that is a course of action that we must seriously consider, we need to be doing the sums to find out how we can make it work.
The biggest worry is that, even with a universal basic income in place, a jobless future will only exacerbate an aching chasm between rich and poor to far exceed the grumbles of the 99% we see now. Whilst those put out of work by automation may be effectively free to pursue creativity, leisure and social lives to their heart’s content, that is as far as the contentment will go when they see how those who were not displaced from regular employment get to live.
Those whose careers persist in spite of, or probably even because of, automation will earn more than ever. This rich elite will have access to healthcare and (yes, it is entirely inevitable) human enhancement that will be unreachable for the unemployed masses.
“AI technologies could exacerbate existing healthcare disparities and create new ones,” argues Russ Altman, Professor at Stanford University in Bioengineering, Genetics, Medicine, Biomedical Data Science and Computer Science. “Unless they are implemented in a way that allows all patients to benefit … A two-tiered system in which only special groups or those who can pay – and not the poor – receive the benefits of advanced decision-making systems would be unjust and unfair … It is the joint responsibility of the government and those who develop the technology to support the research to ensure that AI technologies are distributed equally.”
If we fail to distribute the benefits of artificial intelligence equally, the chasm between rich and poor may come to parallel the evolutionary tale of homo sapiens versus the Neanderthals. Survival of the fittest will be determined by economic success and privilege, with the logical conclusion being that those without the tools to evolve will die – only the post-humans, the new species, will survive.
How can we prevent this? Given this scenario, we cannot reasonably suggest that universal basic income is a workable solution. Should we limit automation in the workplace? Would we even be willing to do so, given the potential financial gains?
Regardless of the downfalls of a solution like universal basic income, we need to be considering how the issue of joblessness due to automation will affect individuals and society as a whole. The ramifications could be extensive and we cannot afford to gloss over the details. An entire era of human civilisation is at stake, and that is far from an understatement.