“What do chatbots, smart refrigerators, payment default risk prediction tools, automatic translators, self-driving cars, email spam filters, and earthquake prediction tools have in common?”
Artificial intelligence resists stable definitions, writes Paola López Merkur (Germany), causing not only conceptual confusion but also concrete legal problems. The term “AI” encompasses a wide range of systems, ranging from generative models to more specific algorithmic tools.
And AI systems aren’t just different in physical form. Technology evolves rapidly, and new versions of models replace older ones at an alarming rate. The result is a moving target that resists clear classification. This “mercury-like” liquidity has significant legal implications. How can something be regulated without first defining “what is regulated”?
Legislative efforts run the risk of being too narrow to capture emerging systems or too broad and lumping fundamentally different technologies together. Additionally, as regulatory debates intensify, companies are intentionally downplaying the use of AI.
PredPol, for example, was the market leader in predictive policing until its home city of Santa Cruz banned the use of such technology. The company changed its name to Geolitica and claimed that it did not offer any predictive tools in the first place. Lopez suggests a similar situation is bound to happen again. “At first, everyone wants to use AI, and everything has AI in it. Because it’s easy to ride the wave of AI hype. But as soon as regulation kicks in…no one wants to have anything to do with AI. ”
Decrease in the value of work
The rise of AI in the workplace doesn’t just endanger workers who perform automatable tasks. It also reshapes the very value of work. Lisa Herzog distinguishes four aspects of the value of work that will be undermined by AI: “didactic, community-building, meaning-making, and political.”
On a didactic level, work is a place to acquire and hone skills. By automating complex tasks, AI limits learning-by-doing opportunities and makes it difficult to gain expertise. This can also affect motivation “if the opportunity to practice and develop certain skills was the very opportunity that attracted the person to a particular profession.”
Second, while work fosters social integration by bringing together people who would otherwise never meet, algorithmic management isolates workers and makes it difficult to develop a sense of a shared culture.
The third dimension of work is its meaning. “Human behavior is structurally ambiguous,” and smaller, more tedious tasks become valuable when they are known to contribute to broader goals. AI-managed platforms take away this meaning from work by outsourcing many small tasks to temporary workers who don’t know their ultimate purpose. Workers begin to feel more like overcoming an “obstacle course of small, complex hurdles” than actually accomplishing something of value.
Finally, the workplace is an “important site of politicization.” They enable conversations about working conditions and workers’ rights and promote political awareness and action. AI negates this aspect of work by reconfiguring the workforce into discrete tasks and pitting zero-hour workers against each other for jobs.
new infinity
Economics, writes Birger P. Pridat, is “the attempt to overcome finite resources by increasing the possibilities of access.” Economic history can be understood as a series of “disciplinary regimes” that begin with fields in the literal sense of the word and expand into different dimensions.
First, the transition from horizontal farming to vertical mining marked a transition from livestock farming to the extraction of finite resources. World trade then allowed European economies to expand their resource base without intensifying production. With the borderless and pathless sea as a new geometrical field, the ship functioned as a vector in the “spatial appropriation of the fruits of foreign continents.” The next area was the temporal area. The industrial economy was driven by investment in future profits and a shift from seasonal growth cycles to constant productivity.
In the 21st century, as the world’s physical resources are depleted, the field has expanded internally, to human behavior itself. “Just as Locke defined Indigenous land as ‘vacant land,’ Google, Meta, and others define our personal data… as ‘raw’ and ‘unowned’ until it is processed by algorithms.” ”
And what will the future hold? Pridat suggests that the sixth field will be biological. We have reached the point of no return on climate. No amount of carbon capture can restore the systems we have destroyed. The only option left for us is to harness the power of AI to reshape the world: solar geoengineering, heat-resistant corals, plastic-eating bacteria, lab-grown proteins, and more.
“The new infinity lies not in the expansion of space, but in the density of design. We are not at the end of the history of productivity, but at the most dangerous and productive point, the transition from unconscious destruction to conscious composition of planetary life.”
artificial personification
LLMs become increasingly human-like, producing responses that feel conversational, empathetic, and self-aware. This anthropomorphic quality is not an accident, writes Max Beck, but is introduced systematically through a series of design choices. For example, the decision to present interactions in the form of “chat” rather than “node-based workflows or command-line tools” is also a deliberate design choice, as is “displaying generated tokens in the chat interface as flowing text reminiscent of human typing.”
The process of creating an LLM begins with a base model trained on a large text corpus to produce statistically valid language. At this stage, “the format of the response is determined purely probabilistically from disparate training data that is not necessarily conversational.” Fine-tuning is then done to adapt the model to more specific tasks, improving relevance and fluency.
The next step is “reinforcement learning from human feedback” (RLHF), where human raters rank and compare the outputs and reward those that appear helpful, polite, or friendly. This process gives the model “personality” by embedding human communication norms into the responses. The result is often a style that mimics emotional awareness.
This artificial anthropomorphism has clear economic advantages and is unlikely to disappear any time soon. Systems that are more human-like are easier to use, more comfortable, have more “stickiness” and longer interaction times. “Ultimately, usage time is the currency of all interactive platforms.” Anthropomorphism is therefore a strategy that aligns user experience with the economic priorities of AI developers and operators.
Review by Cadenza Academic Translations
Source: Eurozine – www.eurozine.com
