Recently I have read a blog post by Jamie Zawinski on Anthony Moser’s opinion about the current developments in AI. Now I want to try to formulate my own arguments against the overwhelming reliance on AI nowadays. It’s been my point of view for a while, however I would like to now clearly state why I think the direction the technology world is heading is wrong.
AI, although currently being hyped beyond reason, has been around since the previous century. However, with the release of ChatGPT to the public, generative models have entered the lives of everyone. As a Computer Science student I have first hand witnessed the effects of a paradigm shift in many domains, and after 2 years I believe that relying on content generated by artificial intelligence is simply harmful.
As an avid fan of english literature I really like reading well-written books. It is a great feeling to be able to appreciate the intricacies of the language and the craftsmanship of the author, who has taken the time (sometimes decades) to write about a certain topic. If you read a lot, you can often tell a well-written book from a poorly constructed one, and if enough people realize this, the society awards great writers with prizes and honors. However, with the rise of large language models, essays, books, novels and much more can be created with a single prompt to the model. While the quality of such writing can often be questionable, it’s important to realize that this takes away the very essence and purpose of writing in the first place. When you put pen to paper you both try to advance your own thinking and convey your feelings and views to a broader audience. It is your opinion and findings that matter, and this is by no means a trivial process. Using artificial intelligence to write for you, or help you write, or correct your writing defeats the purpose of writing something in the first place. This is also the right moment to point out the current concerns regarding this for the book authors and artist of any other kind as well. AI is slowly getting better and better at this kind of work, rendering virtually impossible for me right now to distinguish e.g., electronic music generated by AI and created by humans. This poses a threat to the literature and artistic community, and by proxy, to readers and everyone interested in art. I consciously cannot use such technology knowing that it displaces the very people I admire the work of.
What is even more interesting is that many large language models are trained on books, which are later completely discarded and thrown out. Anthropic, the company behind the Claude AI model, has destroyed millions of print books to train their AI. Here is a very good article about this. In essence, to train the AI, one must scan the books first, preferably quickly. According to Anthropic, the most efficient way to go about this is to strip the books of their cover, rip out the pages and scan just the printed paper. This irreversibly destroys the books, which are later thrown out. It’s a good moment to ask oneself – is this what I’d like to happen to my book, if I ever wrote one? I will not raise the ethics concerns behind such actions, it’s also not my aim to start a debate about this. However, I think the question above is worth asking to yourself.
I think the point made by Hayao Miyazaki, the studio Ghibli founder behind some of the best animated movies of the last century summarizes it pretty well. Recently there has been a viral video going on of him saying in 2016 how he believes AI to be an insult to life itself. As strong of an opinion as it is, I sympathize with his standpoint of view. Being an artist and designer, seeing your life’s work being completely overtaken by soulless software must be terrifying.