In the past, I have struggled to find a lot of necessity for blockchain but I wonder if ChatGPT has actually created a new needed use case- some way to understand provenance?
I’m a person who often likes to read the source paper or to mine the bibliography section of a book to find ways to go deeper on a certain topic or idea. Now, there were already authors who synthesized thoughts and then wrote their own version without attribution, and I suppose it’s fine (cough, cough), and fine for ChatGPT to do its own “remixing” (to use the charitable perspective of aughts-era copyleft), but it also means an explosion of content that may or may not have any reason or careful thought behind it.
In theory, you might be able to train a model with your own curation rules and at least prevent yourself from accidentally incorporating pure nonsense in the results of queries, but then you’d need to evaluate sources (I mean books, writers, possibly publications) and few of us have read enough to cover our bases well here. I read more than the average person and yet every day I can discover many other things I haven’t read by actual real people. To be fair, many of these are just the same content with mildly different perspectives, but there are reams of subject areas that overlap and I have no idea how to evaluate. In science, there’s the added challenge of new developments happening all the time that update and shift what we understand- but can we be sure that ChatGPT can update and capture the nuance of that shift?
I’m not necessarily arguing that people will be better than AI in synthesizing and analysing information in the long run. People do have lived experience that shapes their own personal worldview, and often being able to triangulate my own experience, theirs, and their ideas helps to contextualize information.
But in the short term, we’re about to experience a potential volume of content that, despite being rather prolific themselves, humans can’t produce, to an exponential degree. In the short term, this volume will most likely be in service to selling things, because at the moment, that’s what most of our technology is for- either a metacognative selling in the form of ad-tech or just direct sales of the digital product or service itself. (Social media, like TV before it, is fundamentally ad-tech).
I’m going back to one of my earlier questions – is technology going to start doing the labour technology has created? Is there a level of recursiveness coming that essentially frees me from the digital experience entirely? Could we be headed to a world where one just spends time in person with other people without phones or anything because we’ve replaced ourselves with AI, which is busily selling things to itself with cryptocurrency that doesn’t even apply to physical space at all?
Or, at the very least, will we need to use some kind of immutable ledger to show that a human was involved in ideas and analysis and curation? Because without that, where will attribution exist?
Back when I was filming for Acceleration, I talked to many smart people about their vision of an AI-driven future (or nightmare, as the case may be). That was 10 years ago now, and the way technology developed has still come as a kind of surprise, not that what we have now wasn’t predictable, but how it feels to live in the world where almost all structures of human trust have been torpedoed. That we could be in a situation where simply knowing if something has an individual human source could be the destabilizing force that pushes us over the edge, rendering our main evolutionary advantage (cooperation) moot. On the other hand, maybe we’ll band together in the face of non-humanity becoming a force of dominance. I am living into the latter future.