Written by Hithesh Shaji
McLuhan viewed technology as extension of human capabilities. It centred around human beings and their relationship with technology. The essence of McKenna’s novelty theory explores the universe as an ever-evolving complex system with increasing levels of novelty over time. Novelty can be observed in a wide range of phenomena, including the evolution of life on Earth, the development of human culture, and the emergence of new technologies. In this perspective, biological development of complexity could be the platform or stepping stone for inorganic development of complexity. Both are useful models to understand LLM from a lower level abstraction that is anthroprocentric and a higher level abstraction.
McLuhan’s anthropocentric approach can help us understand how humans shape and are shaped by technology, while McKenna’s perspective can help us understand how technology fits into the larger context of the evolution of the universe. By combining these two approaches, we can develop a more nuanced and comprehensive understanding of the role of technology in society.
A medium in the context of McLuhan’s work is any technology that facilitates communication of information between people or across time and space. The information it transmits could be representations. Bret Victor uses the term “representations” to refer to the different ways in which information can be presented or visualized in a medium. A lot of the power in an idea lies in the form in which it’s expressed because that is what it allows people to think it. Ideas live in representations and representations live in media.
Victor’s criteria for a dynamic medium is that it should hosts artifacts that are
Victor points out dynamic mediums now exists but we are merely emulating and extending static representation from the era of paper. I think LLMs offer us a dynamic medium in which we can represent and transmit information in new ways.
Computational
LLMs are already being used to embed information from a database into a vector space and then used to perform computations on it like searching, classifying and visualising. This is a crude simulation in which the information and the relationship between it is capable of computation.
Responsive
The output of LLMs are based on the prompts (stimuli) or system messages that is defined. These prompts can take various forms, such as natural language sentences, keywords, structured queries and more.
Connected
Initiatives like OpenAI GPTs, AutoGPT, BabyAGI are making LLMs into a medium that can exchange information with apps, database and other AI agents.
LLMs could introduce novel methods of generating, accessing, and distributing their contents, such as utilizing AI agents for networking or sharing personal worldviews through embeddings.
The written word is an extension of our ability to communicate through speech, allowing us to transmit information over time and space in a way that was not possible before. The telephone is an extension of our ability to communicate through sound, allowing us to speak with people who are not physically present. Television is an extension of our ability to see, providing us with a visual window on the world that we might not otherwise have access to.
Every extension of ourselves creates a new environment, or what McLuhan called a “media environment.”
For example, the invention of the printing press in the 15th century extended our ability to transmit and preserve knowledge, creating a new media environment that facilitated the spread of ideas and the development of modern science, philosophy, and politics. Similarly, the invention of the internet in the 20th century extended our ability to communicate and access information, creating a new media environment that has transformed the way we work, learn, and socialize. So, what kind of media environment does the invention of LLM create?
First, LLMs can be seen as an extension of our ability to process language. Like other media technologies, LLMs extend our cognitive capabilities, allowing us to process and analyze vast amounts of language data in ways that would be impossible without technology.
Second, LLMs create a new media environment. This environment is characterized by the ability to process and analyze language data at high speeds, creating new opportunities for communication, collaboration, and innovation. At the same time, this environment can also be characterized by information overload, where the sheer volume of data can overwhelm our cognitive abilities and make it difficult to discern what is important or relevant.
Third, the impact of LLMs on our perceptions and experiences is complex and multifaceted. On the one hand, LLMs have the potential to democratize access to information, breaking down language barriers and making knowledge more accessible to a wider audience. On the other hand, LLMs can also reinforce existing biases and inequalities, perpetuating stereotypes and reinforcing cultural norms that are encoded in the language data used to train these models.
The use of LLMs in mass communication can also create new forms of manipulation and propaganda, blurring the lines between truth and fiction and undermining democratic institutions. LLMs have the potential to facilitate global dialogue and understanding. By breaking down linguistic barriers and allowing people to communicate across cultural and geographic boundaries, LLMs can help to create a more connected and inclusive world. Many dialogues happening in the world have nuances and complexity that can be parsed by LLMs.