Intelligence Augmentation – Agenda for Artificial Intelligence
When discussing the potential risks of work automation through AI, one often hears concerns about the possibility of human replacement. While some may see this as positive, it can also reignite fears of job loss among the general population. This fear arises from the idea that machines are becoming more intelligent than humans — a concept known as the technology singularity.
In this article, I propose that this belief is detrimental to current research in artificial intelligence. While complete human replacement remains a myth rather than a reality, emerging technologies are undoubtedly changing the way we work and presenting new challenges in managing human-machine interactions. These include issues such as work alienation or decision-making power, which require further consideration. My stance is that attention should be focused on augmentation technologies that enhance human capabilities rather than replacing or reducing them. I believe that adopting this approach can bring significant benefits to industries by leveraging the complementary nature of human-machine collaboration.
Human Replacement Is Not A Threat
To better understand the general skepticism towards singularity, it is helpful to distinguish between narrow artificial intelligence (NAI) and artificial general intelligence (AGI):
• NAI focuses on efficiently solving complex problems
• AGI aims to replicate human intelligence
Research efforts primarily focus on NAI, while AGI is still in its early stages and relies more on imagination than reliable data. At this point, there is no concrete evidence of the singularity's existence, and even if it does exist, it is unlikely to occur in the near future.
Singularity is highly speculative in nature. It has sparked considerable discussion and even given rise to what some call Apocalyptic AI. The idea of human self-deification, manifested in the pursuit of a perfect and immortal being, lies at the heart of this modern myth. Interestingly, it is primarily fueled by experts in relevant fields and continues to heavily influence both scientific research and popular perception of AI. A prevalent aspect of this myth is the belief in complete autonomy, which posits that robots and AI will one day function entirely independently. However, this notion overlooks the inherent need for human-defined intentions, making complete replacement impossible.
Despite being a topic of public and scientific discourse, the idea of complete autonomy remains prevalent. This was highlighted in a 2017 Eurobarometer survey, which indicated that 72% of Europeans believe that "robots and AI steal people's jobs." Additionally, a study in the same year predicted that 47% of jobs in the USA were at risk of automation. Another study argued that only 9% of jobs could potentially be automated, but not necessarily in an economically feasible manner. It should be noted that these studies differed in their approaches, one focusing on specific tasks while neglecting their place within larger job roles. Despite previous waves of industrialization automating physical labor, there are still tasks that are beyond the capabilities of current technology. In a 2020 report by the European Commission, the impact of automation on European jobs from 1995 to 2015 was examined, ultimately concluding that it had a positive effect on employment in manufacturing. It is worth mentioning that there are varying perspectives on job replacement due to differing methodologies and levels of analysis. A general survey revealed more optimistic predictions when considering the challenges associated with automation, similar to those seen during past technological revolutions.
Human-Machine Interaction and Conjoint Agency
New automation technologies will undoubtedly change the way we work. They possess advanced data processing capabilities, machines are now able to make decisions and take action on their own, challenging the traditional role of human agency and autonomy in the workforce. This calls for a closer examination of how technology can either empower or mechanize work through a concept known as conjoint agency — the shared capacity between humans and nonhumans to exercise intentionality. There are four forms of this agency:
assisting
augmenting
arresting
automating
They depend on whether it is the human or technology who has control over action selection. I believe that when humans maintain control over action selection, interacting with machines can empower them. However, if technology takes over this responsibility, it can lead to human mechanization in the workplace. Let’s explore the impact of these interactions, and relations between humans and technology to form an understanding of augmented intelligence.
Augmented intelligence
Augmented Intelligence involves AI technology that does not replace human decision-making, but rather provides additional information for decision-support. This falls under the category of Empowerment. For example, imagine an AI software sorting through thousands of pages of text and presenting relevant information to a human decision-maker with a broader view and more context. This process would take days for a human to complete and the machine can only consider a narrow perspective on the data. Augmented Intelligence enables faster and more efficient human decision-making. However, there are challenges when implementing such technology. These include potential discriminatory decisions and employees not being able to contest them due to lack of explainability, resulting in a loss of autonomy and job control.
The use of AI-based worker management (AIWM) systems can be perceived as empowering for managers, but may result in a form of mechanization for workers. This highlights the importance of addressing power imbalances that can be perpetuated by new technology. Discrimination can also be amplified by algorithmic and augmenting systems, leading to unfair decisions for certain groups or exclusion from the labor market. In order to navigate these challenges, it is essential to approach data and augmented intelligence with sensitivity and reflection, considering the moral values that we want to shape human-machine interactions with. When decisions are made by machines, it is important to have specific evaluations in place that require human input to assess fairness and maintain ethical integrity. Ultimately, moral considerations play a crucial role in determining what forms of discrimination, marginalization, and exclusion are permissible, and this decision ultimately rests with humans.
Augmented worker
The Augmented Worker, also known as Operator 4.0, takes a human-centered approach to manufacturing. Rather than replacing workers, the machine is seen as a means to empower them. This marks a shift in the industrial world's mindset — while there has been a focus on replacing humans for years, it is now recognized that they are still vital in factories for the foreseeable future. Some argue that companies can achieve even greater success by utilizing the complementary strengths of humans and machines: the flexibility of human work combined with the efficiency of automation. However, in reality, most manufacturing systems continue to operate under a traditional technocentric mindset where technology dictates the work process.
The primary focus is on performance and repeatability, with the expectation that humans will function like machines. This may lead to a sense of mechanization. Recent studies have highlighted the potential for further division in the job market: those who serve technology (the cyberproletariat) and those who possess the skills to control it. But with the rise of technologies like collaborative robotics, machines are becoming more accessible and affordable for non-experts. These advancements, along with AI, can be used as tools to support and enhance workers' abilities rather than restricting their decision-making to prevent mistakes. By empowering workers on the shop floor and giving them control over new technologies, productivity, motivation, and innovation can all be boosted to their maximum potential.
Conclusion
The fascination with the idea of a singularity surpassing human capabilities remains strong among both researchers and practitioners. My argument counters the common belief that singularity is the main concern. Instead, I believe we should shift the focus to the issues that shape our daily interactions with machines. This means moving away from the debate on human replacement and towards examining the quality of human-machine interaction. By studying the complexities of this joint agency, two types of interactions:
those that mechanize humans
those that empower humans
To build a positive future, our attention should be on promoting empowering interactions. In pursuit of this aim, the agenda should include considerations such as fairness, discrimination, skill, and power within organizations. It is important to note that it is not technology itself that poses a threat to humans, but rather its implementation. Therefore, when designing technology, we must prioritize questions related to agency and power in order to adopt a human-centered approach and create sustainable employment and learning opportunities for all individuals.
Title image credit: Michel Poisson