A brief history of the Web could read ‘In the service of profit, everything open was fenced in’, which sounds like an opening line from a Cormac McCarthy novel. The result is platform capitalism or vast walled gardens of data and activity. So vast we often forget they have edges.

AI follows this model, hoovering up all available data into an inscrutable set of probabilities shrink wrapped in language. This can be useful for ‘evergreen’ content but is limited when we want to go beyond anything which has gone before.

Whether it’s ‘classic’ search or an LLM, it is easy to fall into the idea that these ‘places’ contain all that can be known. That all answers are available because everything that it is possible to know is online. We assume the garden is so vast that the walls cease to matter. Of course, this is the impression that any large platform wants you to have, ‘there is no need to wander off’.

Hyponiscience

In effect we treat many platforms as if they were all knowing, or omniscient and thereby put ourselves into a state of Hyponisience. A false sense of epistemic mastery.

For most of the time this is relatively harmless. Most information seeking is within a standard canon; there are correct answers. However, when we are looking to produce new knowledge and new thinking, or even to expand our worldview Hyponisience becomes problematic.

While the emergence of AI is what led me to invent this term, it is not a new problem. It also describes the process of being algorithmically crammed into an ever decreasing epistemic space driven by an attention economy. Hyponisience is also a state of only believing the information within your bubble. A state which fuels polarisation and which is form many a haven in the face of the complexities and pluralism of information abundance.

How do we counter this false state?

All flavours of information literacy advise engagement with multiple sources. Now perhaps we should advise to engage with multiple platforms online, and occasionally offline. Rather than a hierarchy of quality (From Journal Papers through to Overheard in the Pub), we could have a quality of range: how many places did you draw information from? How distinct were these places? Reaching beyond the walls of a single garden and having a wander can only lead to a broader, more meaningful, view. This also opens up Post-Critical possibilities, whereby some of the ‘places’ we seek might be people and/or embodied knowledge.

For example, as I outlined in my ‘AI Learning Gambit’ post, a new literacy involves knowing when not to use, or to go beyond, AI. In a post-provenance era, the best way to maintain some agency and see beyond the walls is to actively choose the less convenient, but possibly more rewarding, options.