By Uma Rani, Sr Economist at the ILO and Rishabh Kumar Dhir Technical Officer at the ILO
While AI is celebrated as a driver of automation, its success hinges on an invisible workforce performing low-paid, precarious tasks under challenging conditions. This article unpacks the hidden realities of AI’s “human-in-the-loop” model and its profound implications for workers and society.
Artificial intelligence (AI) is often presented as a revolutionary force poised to automate vast swathes of the economy, displacing workers, and ushering in a “post-work” era. However, behind the sleek interfaces and impressive capabilities of many AI systems lies a hidden workforce of humans. This “human-in-the-loop” model reveals a more complex reality, one where AI is less about replacing humans and more about relying on workers with decent work deficits, such as low earnings, lack of social protection benefits and occupational safety and health to sustain the AI system. This is what we look at in our AI-enabled business model and human-in-the-loop (deceptive AI) article, which examines how these workers power automated systems and the implications for labour markets, society, and for the workers themselves.
Invisible labour in the development and deployment of AI
From self-driving cars to virtual assistants, the AI industry thrives on data. This data needs to be meticulously labelled, categorised, and annotated. This requires human intelligence and labour – both of which still cannot be replaced by machines. Such tasks are often outsourced to crowdworkers on digital labour platforms or to Artificial Intelligence-Business Process Outsourcing (AI-BPO) companies. These platforms fragment complex tasks into microtasks and offer small payments for each completed task. Crowdworkers, whom are also known as invisible workers because they often work behind the scenes, are essential for training AI algorithms on several functions, such as text prediction and recognition of objects.
Similarly, virtual assistants, marketed as autonomous tools, often rely on invisible workers who may be transcribing audio, verifying the virtual assistant’s understanding, or even performing tasks like scheduling meetings that AI may struggle with. Even sophisticated large language models with impressive capabilities rely heavily on human trainers to fine-tune their responses and mitigate biases, toxicity, and disturbing content. As a result, workers are routinely exposed to graphic violence, hate speech, child exploitation and other objectionable material. Such constant exposure can take a toll on their mental health and trigger post-traumatic stress disorder, depression, and reduced ability to feel empathy.
Global Digital Labour: What are the opportunities and challenges?
Despite this, platforms offering microtasks and other AI-related work continue to find willing workers because they often provide opportunities for remote work, offering flexibility and accessibility for individuals in different locations and circumstances, particularly in developing countries or with limited access to traditional employment. This can be a crucial source of income especially for those facing barriers to traditional labour markets.
An ILO survey of crowdworkers reveals that many of them are highly educated, holding bachelor’s or postgraduate degrees, often in specialized fields like science, technology, engineering, and mathematics (STEM). Yet such workers are primarily employed in routine and repetitive data work, which often require minimal specialized knowledge. This leads to a significant mismatch between the educational attainment of workers and the AI-related tasks that they are hired to do.
Workers are therefore performing tasks that do not leverage their educational background or offer opportunities for intellectual stimulation or professional growth. This leads to job dissatisfaction and insecurity among workers, and their underutilization represents a lost opportunity for leveraging skilled labour to drive economic growth and innovation for developing countries who commit significant resources to invest in higher education.
The survey also found that the median earnings of these workers in developing countries are about US$2 per hour, and they have limited social protection and a high risk of work rejection due to automated decisions that occur, without clear justification or a communication channel to address these issues or any worker grievances.
The implications for the future of work
The reliance on human labour in the AI industry raises several critical questions. Firstly, there is a risk of deskilling workers and hindering their career development. Second, the shift towards contingent work arrangements can contribute to lower labour share in income, and it can increase income inequality.
The human-in-the-loop crowdworker model underscores the need for a more nuanced understanding of what is often believed to be “automated” and the impact of AI on the labour market. While AI holds the potential to enhance productivity and create new opportunities, it is crucial to address the ethical and social implications that arise from it, including the need to protect workers and promote greater transparency and accountability in AI systems.
Initiatives by governments, social partners, and other stakeholders
Governments, social partners, and other stakeholders are increasingly recognizing both the opportunities and challenges that AI systems present for the future of work. In the United States, an executive order emphasises worker and employer involvement in AI policy development and implementation. The European Union is pursuing similar goals with its AI Act and digitalization framework agreement.
Best practices and codes of conduct are also emerging, including independent initiatives like the Fairwork AI Principles. Certain crowdwork platforms have adopted the “Crowdwork Code of Conduct” to improve working conditions, and some companies are offering wellbeing services for workers who review disturbing content.
While these developments signal growing recognition of the challenges within the AI supply chain, more is needed, particularly in developing countries. Social dialogue, where workers are visible and in conversation with employers and governments, must be at the heart of such processes if we are to ensure that the benefits of AI are shared equitably, and risks are mitigated.
The future of work in the age of AI should be one of genuine collaboration between humans and machines, not one built on a hidden, global workforce facing decent work deficits. Only then can we realize the full potential of AI while ensuring a more equitable and sustainable future for all.
