Is an Epistemological Crisis Looming with the Advancement of ChatGPT and Language Model AI?
The recent advancements in language model AI, such as ChatGPT and GPT-3.5, have sparked a new wave of discussions and debates in the technological world. With these AI models being trained mostly on material written by humans, especially from the web, the web is now filling up with text written by AI. This raises the question of whether or not AI’s own output will soon infect the content of academic papers, books, and other sources of knowledge.
The increasing prevalence of AI-generated text has led many to question the future of knowledge and the impact it will have on society. With future AI being affected by its own output, the purity of knowledge may become contaminated. It’s like the idea of animals eating their own and developing prion diseases.
However, some argue that this new era in AI might lead to an epistemological revolution, forcing people to have a better relationship with truth and facts. Religion, on the other hand, might face a hard time in this new environment and may see itself as an AI-free zone, but it is unlikely to last long.
No part of the above was written by AI, but was part of an email reply I sent to someone who was shocked by the excellence of the replies that ChatGPT gave regarding some problems of economics theory. As I tend to write emails in all lower case, as was the habit thirty years ago, and it is so much faster that i have never seen reason to change, it does mean that when I want to base an article on something I have written, I have to go through and change all the cases or get an algorithm to do it, mostly very imperfectly. However, ChatGPT does a perfect job of re-forming such text, so I fed my email into it and asked it to fix it up for a blog post. To my great pleasure, not only did it instantly capitalised everything perfectly, it also broke it up into paragraphs (often a tricky task, at least for me) and suggested a title, which I have used. The title is pretty close to my first sentence and is not exactly red hot, but it will do.
However, being a chatty AI chat-bot, ChatGPT could not resist adding its usual little coda, which I think OpenAI’s lawyers have encouraged to bolt on to practically everything, like a valetudinarian customer care representative saying ‘Have a nice day’. By the way, what is going to happen to customer care reps and especially all those call centres? Anyway, here’s the coda:
In conclusion, the advancement of language model AI presents both challenges and opportunities. It is up to us as a society to navigate this new landscape in a way that promotes knowledge and truth.
A nice cheeful and optimistic Hollywood ending, and most appropriate given that society has already done such a fabulous job of navigating the noetic landscape in way that promotes knowledge and truth, as any follower of Donald Trump or tele-evangelist or – well, you can fill in the gap – will tell you.