IBM recently unveiled Project Debater, a program that uses artificial learning to provide arguments in real time on any given resolution. For this week’s Editors’ Forum, program editors and contributors of the NATO Association of Canada comment on the increasingly fast development of artificial intelligence and the expansion of its application.
Adam Zivo, Junior Research Fellow
Recent advancements in artificial intelligence are less impressive than they seem. The newest generation of AIs can best be described as glorified statistical analysis. While these AIs may be able to complete increasingly complex tasks, they lack consciousness and the capacity for abstract thinking. The Chinese Room thought experiment, which has been a staple in the philosophy of mind for decades, demonstrates this fairly well. Rather than thinking of our AIs as minds, we ought to think of them as remarkably effective, task-oriented computer programs. This lens calls attention to their fundamental inflexibility and narrowness. Project debater excels within the task it was designed for, but it doesn’t have the capacity to “think” beyond that task.
The overall social implications of AI are mixed. Throw away any science fiction notions of sentient, truly human-like AIs. We don’t even have the theoretical basis for that yet, as the architecture of consciousness remains both enigmatic and contested. Zeroing in on the flavour of AIs we have right now, their role in society will be limited to their ability to automate tasks, an ability that is both their greatest benefit and danger to society. Theoretically, it’s wonderful that we can outsource labour to machines. However, this also risks creating a perpetually terrible labour market for humans. What happens to the millions rendered unemployed by self-driving cars? Or call centre workers made irrelevant by chat bots?
We’ve experienced this problem before, sort of. Advances in technology and automation have always dispensed with some category of worker or another. Globalization, too, created seismic shifts in our labour pool. What’s anxiety-inducing about AI is the *unprecedented scale* with which this would happen. The worry is that we’ll see a hyper-version of the general problems associated with automation. In practice, that looks like a society of two classes: (1) the elite who own the machines which monopolize labour and its profits; and (2) a vast underclass of workers struggling to eek a living by cobbling together low-paying jobs. Can our systems manage the disruption of AI automation and preempt this outcome? That’s genuinely impossible to answer, so I guess we’ll see.
Nasser Haidar, Program Editor: Emerging Security
When it comes to artificial intelligence, as I’m sure the Project Debater machine created by IBM can tell you, it is not a straightforward topic. It’s no secret that millions of jobs that can be easily replaced by automation and AI will be at risk, once it’s rolled out as a fundamental aspect of our global workforce. As a result, governments must be proactive with their progressive policies, making space for AI’s benefits within our economy, and also mitigating the effects of a massively underemployed population.
I don’t agree that it will be a bane to society, nor do I blindly believe that it will simply be a boon to society. Before we reach that stage, we must learn a lot more about Artificial Intelligence than it does about us, incorporating as much regulation as we can in its development., guiding its continuation through a narrow path from this very day forward.
Nevertheless, when it comes to what we’ve seen in regards to Project Debater, there is a slightly different aspect of the AI debate we’ve not yet experienced. Project Debater can put together complex ideas and arguments based off of hundreds of pre-downloaded academic sources. If it can do so, then the worry of them putting humans out of work seems minuscule when compared to its ability to replace human thought altogether. Does this mean that when Artificial Intelligence is at an advanced stage of development, education simply becomes futile? Not to forget the possible deadly military application of AI as well.
Of course, it’s not as simple as, let’s say, creating a ‘space force’, but Project Debater has shown that it is not without reason to be worried about the convergence of AI’s abilities to replicate decades worth of human academic, emotional and cultural capital. If IBM had gone out of its way to emphasize that Project Debater was not connected to the internet, and its knowledge comes from pre-selected sources, does that then not beg to ask the question: What will happen when it is’s finally connected to the internet? And even more importantly, is it reversible? Only time will tell.
Maria Konstantina Pepelassis, Program Editor: Security, Trade, and the Economy
Artificial Intelligence is quickly becoming more science than fiction. Project Debater proves this, and furthermore proves that what is currently understood as a uniquely human capacity is quickly capable of being artificially reproduced in robots. This will prove to become hugely problematic in our society. These developments do not replace simple human functions like sewing or painting, as experienced in previous industrial revolutions, but may prove to replace functions deemed fundamental to human purpose, including occupations that require the use of reason or the practice of creative skills. Project Debater shows reason and argumentation are no longer uniquely human traits, but a skill reproducible in robots.
The issue with this reproduction of what were believed to be uniquely human traits is the fact that the skills we have will no longer be employable to a degree beyond simple human labour. As the list of skills we can use to make a living shrinks, so do our prospects for work and a source of income. The McKinsey Global Institute found that of 750 jobs, “45% of paid activities could be automated using ‘currently demonstrated technologies’[MP2].” While currently many of these projections show blue collar jobs face the brunt of automation, Project Debater shows more could be at risk in the near future.
Understanding the impact of AI on our job sector, there are obvious threats to the international rules-based order. Any threats posed by AI to jobs will most likely translate into isolationist political policies with the aim of ensuring the continued security of those jobs at risk. People will be more focused on ensuring peace and prosperity at home than they will on an international level. While studies on the impact of AI are still in preliminary stages, the isolationist policy of the United States following WWI that coincided with the Great Depression shows the validity of this perspective. Once the economy began to turn in the 1920s, American foreign policy was no longer deemed a priority. Indeed, extreme restrictions were placed on immigration, tariffs were imposed on foreign goods to shield manufacturers, and America remained neutral during WWII until targeted by German submarines. All of this was done to ensure American interests and citizens were protected from the interests of alien states.
Project Debate signifies not only the technological advances that will propel efficiency and productivity, but also the end of employment and life as we know it. Whether we will be able to establish the social systems that will enable people to live comfortably as they have before remains to be seen. What is reasonably certain is that if we don’t, the organizations and institutions we can afford to support as wealthy nations with prosperous citizens are in danger.
Amaliah Reiskind, Program Editor: Cyber Security and Information Warfare
Project Debater goes deep into the uncanny valley, and is therefore garnering a lot of attention. The technological developments that let the AI present appropriate and real-time arguments with a human cadence are impressive. It is important to remember though that it is still far removed from being a “thinking” being. Its responses are based on rote training, algorithms, and access to “several hundred million” articles. This is evident in some of its arguments where it talks about related topics but doesn’t directly address its opponent’s point. Project Debater is a “narrow” AI that can find correlations and present them, but doesn’t understand the full context.
It can sometimes be forgotten when AI applications seem to predict our needs, but the creation of an actual “thinking” AI is still far away. The majority of jobs will continue to need a human’s ability to think abstractly, or need human oversight. So yes, the continued development of AI will probably shift the job market a little, but it will overall result in easier lives.
If and when an actual thinking AI is created it will far surpass our cognitive abilities, and then we should definitely worry. Until then there are many other things we should be concerned about. One of the reasons AI has been developing so fast in recent years is the crazyastonishing amount of processable data available for AI input. While it has become apparent that your personal data is a commodity resource, it is still a bit of a Wild West when it comes to legislation regulating the use of it. The GDPR in the European Union has taken goodpositive steps to check the use of its citizens’ data, but many other countries are still lacking. The idea of AI would be a lot less scary if you knew what data was being used to create its “brain”. It’s therefore time to start paying attention to the man behind the curtain. We can help ensure AI grows in a way that will continue to augment our society, instead of taking over it; data governance needs to solidifybe implemented, development needs to be monitored, and global dialogues and platforms should be fostered. This way we can aim for a more Star Trek future than a Matrix downfall.
Photo: Artificial Intelligence, by Seanbatty via Pixabay. Licensed under CC0 1.0 Universal.
Disclaimer: Any views or opinions expressed in articles are solely those of the authors and do not necessarily represent the views of the NATO Association of Canada.