Tag: AI

AI Steals My Words

I’m tempted to have AI blog for me today, because I’m tired from lack of sleep. But I would never do that, because I know what generative AI is: a plagiarism of what’s available on the Internet.

Artificial intelligences such as Chat GPT are “trained” on Internet content. That means the AI studies composition, word usage, style, and content. It captures the writing itself and uses it in other combinations for its own work. What makes my writing unique is my choice of usage, style, composition, and content.

I’m a writer. I don’t like that generative AI can take my work and make it theirs. It seems like an appropriation of my creativity and that of others. I especially don’t like what it does to visual artists, because stealing pieces of images seems more blatant than just stealing words.

Somewhere, an artificial intelligence is scanning this and putting the information in with other writings it’s scanned. And maybe it will spit it out verbatim into someone else’s writing. I don’t know, and maybe that’s the worst part.

NaNoWriMo and Generative AI

The controversy in the writing world currently is that the NaNoWriMo organization has issued a statement not only supporting use of generative AI in its events, but dismissing opposing viewpoints as ‘ableist’ and ‘classist’.

To understand the impact of this, let me start with NaNoWriMo. This organization sponsors a world-wide writing festival every November which encourages people to write a 50,000 word novel in one month. Admittedly, 50k is somewhat short for today’s expectations of a novel, but it’s 50k more than most people feel they can write. In 2020, 383,064 people participated in NaNoWriMo, the latest statistic available (Wikiwrimo, 2024). For full disclosure, I have participated in NaNoWriMo for several years.

The issue with generative AI is more complicated. Not all AI is generative AI; that is, not all AI is used to generate or create content. The fear of writers is that generative AI creates content, and it creates it from the materials it’s been trained on, which are existing works. This goes beyond analyzing patterns in grammar use and spelling (which I would argue are acceptable) into creative aspects. In other words, training generative AI is mass plagiarism of ideas without crediting sources. An entity like NaNoWriMo supporting mass plagiarism of ideas seems antithetical to its principles.

In addition, artists and writers fear being replaced by the much cheaper generative AI. The quality of generative AI is not as good as the actual creations of human beings; but if generative AI takes over in commercial outlets, the public will become inured to lower quality. The loss of revenue to real live writers will become the loss of creativity to the wider world.

To address NaNoWriMo’s charge that opposing their approval of generative AI is classist and ableist, it is classist and ableist to assume that people with disabilities or of underrepresented social classes would need to use generative AI to compete in the marketplace of ideas. I suspect that the issue here is a lack of distinction between AI used to proofread and suggest grammar (such as in ProWritingAid, one of their sponsors) and the AI that creates entire segments or whole stories. I see a big difference between supporting a tool for improving form and a system for writing content. If this is NaNoWriMo’s dilemma, then they need to do some soul-searching and make a clear ethical statement as to where the line gets drawn between composition tools and content creation.

This is where I am in my ethical processing of the issue, that use of AI for translation, proofreading, or grammar correction is not at the same level as AI to generate ideas and content. The former is predicated on objective rules; the other on skimming subjective creative works. My struggle to define what is permissable is the struggle of the entire society in dealing with AI.

Wikiwrimo (2024). NaNoWriMo statistics. Available: https://www.wikiwrimo.org/wiki/NaNoWriMo_statistics#cite_note-3 [September 3, 2024].