AI: the Anti-Internet
AI was born from the internet.
It could not have emerged without the fertile ground of tokens and interconnected semantic meaning that the web provided for its training.
Yet, as a technology, it manifests many opposite characteristics.
The internet is, at least conceptually:
Open: all information is publicly accessible.
Free: available without direct cost.
Interconnected: inherently networked and collaborative.
Conversely, Large Language Models (LLMs) represent nearly the inverse:
Closed: the largest and most powerful models are proprietary, their secrets closely guarded.
Metered: usage incurs direct costs.
Engulfing: their value grows with the accumulation of data, but, due to their closed and metered nature, this value accrues disproportionately to their owners and privileged users rather than broadly benefiting everyone.
This divergence has profound implications, impacting everything from technological economics to product development strategies.
Yet, AI does not replace the internet; the two coexist and intermingle.
Increasingly, the internet will be both created and consumed by AI. Early experiments, such as the semi-sentient art project
, foreshadow this evolving landscape.Perhaps, free from the burden of our screens, able to interact with other intelligences (human or artificial) in more multimodal ways, we can become more authentically human in the process.
The early internet was bright and utopian, but it gave way to corporate domination and social media excess.
AI emerges under a dark cloud, with great fear and uncertainty surrounding it.
Perhaps it will continue to follow an inverse path.
I started blogging in 2008. Blogs and newsletters were always free. I never imagined that we'd end up asking people to *pay* for articles on platforms like Substack. And yet, here we are.
Everyone wants what other people offer to be free, while at the same time they want to get paid for what they offer themselves. This doesn't add up.