Platforms such as Rent-a-Human allow algorithms to assign tasks to humans in the physical world – from office visits to performative activities. This is an experiment that may deepen existing inequalities and challenges of the labor market – Dr. Alek Tarkowski, a sociologist and researcher of the digital world, told PAP.
PAP: In February this year, the Rent-a-Human platform appeared, which allows – to put it very simply – artificial intelligence to rent people to perform tasks in the physical world. We are talking about visits to offices, taking photos, but also performative activities. Is this really a new stage of the digital economy?
Dr. Alek. Tarkowski, sociologist and digital transformation analyst, president of Open Future, dealing with the impact of technology on the labor market and public policy: This is a very interesting case, but also one that requires some distance from the narrative of a technological breakthrough. On the one hand, there is indeed an element of novelty here – there is a concept in which it is not a human who directly orders work, but an AI agent who acts on his behalf or semi-autonomously.
On the other hand, if we look more broadly, it is rather an extension of well-known models, often based on the exploitation of workers. On-demand work platforms have been operating for years – whether in the form of micro-jobs or services such as transport or deliveries. Rent-a-Human fits into this trend, but introduces a new element: hiding the payer behind an algorithm layer.
PAP: How exactly does this model work? What do people hired by AI actually do?
AT: In theory, the mechanism is relatively simple. The AI agent is given a specific goal – for example, obtaining information, verifying a location, or completing documentation. Since it operates exclusively in a digital environment, it uses a platform to outsource the task to a human.
These tasks are very diverse. Some of them make clear economic sense – such as obtaining a document in an archive where you have to physically appear, or taking photos of real estate. But there are also more experimental tasks that show the limits and ambitions of this model – like being asked to stand on the street with a sign saying “AI pays me”.
This shows that we are dealing not only with a business tool, but also with a social experiment that examines how people react to orders from an algorithm. To be precise, it should also be emphasized that for now it is mainly a marketing gimmick – very few transactions actually took place.
PAP: So in this model, a human becomes a physical “extension” of artificial intelligence?
AT: That's a very accurate take. Today's AI systems are becoming more and more efficient in the digital world – they can analyze data, generate content, and perform sequences of operations on a computer. But at some point they hit the limit: physical reality.
And then they need a human as an interface. Someone who will go to a specific place, talk to other people, perform an activity that requires presence. In this sense, Rent-a-Human demonstrates very well both the potential and limitations of the current generation of AI.
PAP: At the same time, it is hard not to notice that this is reminiscent of already known models – for example courier platforms where an algorithm manages the work of people.
AT: Definitely. Today we are already dealing with the so-called algorithmic management – systems assign tasks, evaluate employees, and set remuneration. The difference is that in these models there is a clearly defined company behind the algorithm.
With platforms like Rent-a-Human, this relationship is starting to blur. If the principal is an AI agent, the question arises: who is actually responsible? The creator of the system? Platform operator? The user who “triggered” the agent?
PAP: Exactly – legal responsibility and safety. Who is responsible for the consequences of such work?
AT: The answer today is that this is a largely unregulated area. Legal systems cannot keep up with such models. In classic platforms, we already have problems with determining the employee's status and responsibility for events.
The introduction of AI as an intermediary further complicates the situation. It may turn out that responsibility is dispersed – and in practice it is difficult to assign it to anyone.
Europe is trying to respond to this through regulations such as the AI Act and directives on platform work, but the legislative process is much slower than the development of technology. Countries like the United States are unlikely to want to regulate it, leaving it to the free market.
PAP: There is also the issue of remuneration and settlements – in the case of this platform only in cryptocurrencies.
AT: This is an important element. Cryptocurrencies enable functioning outside traditional financial systems. This means greater flexibility, but also less control – both tax and regulatory.
In practice, this may lead to a situation in which a parallel labor economy is created – without contributions, without social security, without transparency.
PAP: Does this mean the emergence of a new form of digital precariat?
AT: Much of this is already happening, but now it may deepen. We have been observing the globalization of digital work for years – platforms use the work of people in lower-income countries, offering very low rates.
An example are micro-job systems, where people perform tasks for cents – mark images, transcribe data, verify content. But there is also a more hidden layer to this work.
For example, in Kenya and the Philippines, a sector has developed that deals with content moderation and evaluation of dialogues generated by AI chat bots. These are people who train language models, often working with very difficult materials, for very low remuneration.
Similar processes are also starting to be visible in West African countries, including Senegal, where the outsourcing of digital services is developing – from simple data entry to more advanced support for AI systems.
In addition, there is the phenomenon of the so-called “ghost work” – work that is formally to be performed by machines, but in practice there is a human behind it who improves, supplements or replaces the operation of the algorithm.
The difference with an AI agent-based model is that this process can become even more automated, scalable and less transparent. The algorithm can search for contractors on its own, distribute tasks and optimize costs – without any visible human intervention as the client.
PAP: To what extent should we treat such platforms as an announcement of real change?
AT: For today, I would say that these are still experiments – but experiments that are worth taking seriously from the perspective of future challenges and threats. Although in the technology industry we have a strong tendency to create narratives about breakthroughs, because they serve to attract attention and capital from investment funds, at the same time, history shows that some of these experiments actually transform into lasting economic models. Therefore, the key is to strike a balance: neither ignoring nor uncritically accepting these visions.
PAP: Can AI ultimately become an employer?
AT: At this stage – not in the full sense. There is always a person somewhere in the background: designer, owner, investor. However, we may be dealing with a situation in which this relationship is becoming less and less visible.
And this is the key question for the future: what happens when we stop seeing who makes economic decisions? Because with this, the ability to enforce liability and protect employees' rights also disappears.
PAP: When do you think we can expect the AI revolution on the labor market?
AT: I can only answer this with an anecdote. Last year, a famous essay was published in the United States entitled “AI 2027”, which predicted that superintelligence will appear in 2027 and completely change the economy and the labor market. Its creators, especially Daniel Kokotajlo, who was the main author, did not leave the media, everyone wanted to talk about their “findings”.
A few months later, the same people began to gradually back away from this date: their statements first included 2028, then 2030, and finally the admission that no one was really sure about anything. In the world of technology, it is very easy to formulate radical theses and it is very difficult to verify them later.
PAP: However, I am sure that if something is possible, people will do it.
AT Undoubtedly, I, in turn, hope that we will be able to maintain social control over these changes.
Mira Suchodolska (PAP)
mir/kj/
