Artificial Intelligence is coming for us. But long before it comes for us, AI is coming for its own progenitor, the internet. Its ability to churn out poetry, allegedly indistinguishable from Shakespeare’s, or its newfound trick of mimicking Hayao Miyazaki’s art, do not justify the paranoia about how AI will affect humanity. However, wariness about AI’s incursions into the world of humans often leads us to overlook how fundamentally it is changing all that is beautiful about the internet.
It has been one year since Google announced its plan to move away from PageRank toward AI summaries last May. In the 1990s, Sergey Brin and Larry Page, the co-founders of Google, had come up with an extraordinary idea: the PageRank algorithm. What it did, in essence, was determine the importance of web pages based on the number and quality of links pointing to them. PageRank was based on the principle outlined by the sociologist, John R. Seeley, who wrote in 1949 that “a person is prestigious if he is endorsed by prestigious people.” The web, Brin and Page showed, is not just a barren lexical environment full of reams of data but a social milieu where affiliations in the form of hyperlinks corresponded to prestige and reliability. But this is not all that a hyperlinked web does. It also presents a map to a wonderfully interconnected world that documents how people’s thoughts naturally move from one idea to another. Thus, hyperlinks illuminate the pathways of thought that lead people to think in certain ways as well as link their thoughts with those of others.
This harks back to an older concept that all ideas are linked to others. It is based on a philosophical theory that originated way before even the lighting of the first bulb in 1879, let alone the invention of the internet. In the 17th century, Thomas Hobbes and John Locke debated the extent to which a person controls the succession of ideas appearing in his/her mind. They posited that trains of thought were the outcome of interactions between the data received from the senses and one’s mental faculties — reason and imagination. David Hume went a step further and argued that all successive ideas are linked by association, outlining three kinds of connections: resemblance, contiguity, and cause and effect.
Now think of how a user researching a topic navigates the web. An initial query yields a list of results and the researcher opens several of the links in new tabs. He/she skims through the pages, clicking on embedded links that point to aspects of the topic to be explored in greater detail. As the researcher follows the different associative trails, he/she confronts varied and, at times, contradictory lines of reasoning. The researcher’s own thoughts begin to take shape as he/she figures out how to resolve the tensions and conflicts that arise across multiple pages. The links, then, matter as much as the text when it comes to thinking.
But now, we stand at a turning point in the internet’s history where the humble
link — the endless tapestry of ideas woven together throughout the web that once was the heartbeat of the internet— is in danger of going extinct. This is because large language models, such as ChatGPT, Google’s Gemini and so on, extract the information represented online and present it in source-less — soulless? — summaries. The work of making connections both among websites and in a person’s own thinking is what AI chatbots are designed to replace.
In sanitising the lively web of hyperlinks, AI is guilty of committing what the literary critic, Cleanth Brooks, once denounced as the “heresy of paraphrase”. Paraphrasing, he argued, is heretical because it flattens the tensions, ironies and conflicts hidden in each line of text. A world wide web stripped of hyperlinks is similarly blasphemous.