Tornado Alley (PROG)

[fuagf]

Reposted: SWEET! "Let's stop calling it "Artificial Intelligence" then and call it for what it is and does a "plagiarism software" because "It doesn't create anything but copies existing works, of existing artists, altering them enough to escape copyright laws.
It's the largest theft of intellectual property on record since European settlers arrived on Native American lands.""

At least that was my initial impulsively emotional response. Then because i don't really understand AI and algorithms nearly as well as many i decided to look into it a bit further. And without raining too much on the quote, if in fact the whole is Chomsky's, i got one connecting it to AI and art. In any case the analysis this guy gives as regards to plagiarism reads to me as legitimate:

Thoughts on AI “plagiarism”

I recently just read a meme on Facebook about AI-generated art and plagiarism. I wasn’t impressed.

Here it is:

---
Let’s stop aggrandizing it by calling it “artificial intelligence,” and begin calling it what it really is: plagiarism software.

There is no artificial intelligence. It’s not creating anything, just copying existing work by existing artists and changing it enough to skirt copyright laws: there’s a reason why everything is all “in the style of Wes Anderson” or “in the style of Ernest Hemingway.” So don’t say “an artist generated these images using artificial intelligence,” but rather “someone made this using plagiarism software.”

Don’t say “a marketing company utilized artificial intelligence to make this ad,” but rather “created by a marketing company using plagiarism software.”

By allowing the companies making the plagiarism software to call it “artificial intelligence” or “text algorithms,” you’re giving them newspeak to hide what they’re making, and what their users are doing: it’s just plagiarism. Nothing else.
---

The author apparently doesn’t know how algorithms work. And the use of the term ‘plagiarism’ is over-extended and misleadingly reductive.

Not to mention how poorly reasoned it is. Fallacies within fallacies:

1. Straw Man Fallacy: The argument presents an exaggerated and oversimplified view of artificial intelligence (AI) by claiming that it is merely copying existing work without any creative capabilities. This misrepresents the concept of AI, which involves the development of algorithms and models that can generate original content based on patterns and data.

2. False Dichotomy: The argument asserts that AI is either creating original work or engaging in plagiarism. It fails to acknowledge that AI-generated content can be a combination of both by using existing data to generate new and unique outputs. Of course, that includes data about patterns people use to create stuff.

3. Appeal to Emotion: The argument uses emotionally charged language, like “aggrandizing,” “hiding,” and “plagiarism,” to evoke negative sentiments towards the use of AI in creative contexts. Emotional appeals do not provide anything substantive to support arguments like this. Emotions in a logical argument are not evidence.

4. Sweeping Generalization: The argument makes a sweeping generalization by assuming that all AI-generated content is plagiarism without providing sufficient evidence to support this claim. It overlooks the diverse applications and capabilities of AI in different creative fields.

I suppose the oversimplification and sad logic are at least a place to start a conversation (although I wouldn’t waste my time with the author).

About patterns… in my graduate-level linguistics class, we were studying Chomsky’s transformational grammar [Ouch .. https://en.wikipedia.org/wiki/Transformational_grammar .. heavy] For my term paper, I did a statistical analysis of several comparable paragraphs from 3 authors and the transformational rules they used. The professor said he’d never seen something like that and posited, way back in 1979, that we could learn to “write like famous authors.”

Enter algorithms that take my statistical analysis much further (and faster).

As for the term “artificial intelligence”, Chomsky had an article in the NYT explaining how what we have now is NOT artificial intelligence. And most scientists in the field agree with him.

However, the reasoning and content-generation capabilities are currently smarter than most humans in the sense that the AI can connect conceptual dots better and faster than most humans. That’s not to say AI doesn’t make mistakes — it does, lots of them. But then, so do humans (spectacularly). For instance, AI cannot reason on morality. Nor can it understand something called implicature, understand sarcasm, and it misses a lot of abstract concepts.

The specific topic of plagiarism, however, is an interesting topic. Plagiarism typically involves the unauthorized use or copying of someone else’s work and presenting it as one’s own. In the case of AI-generated paintings, the AI is trained on existing artworks, including those by renowned artists like Van Gogh, and it generates new artworks based on that training. The AI is not directly copying or replicating specific existing paintings, but rather creating new works that are influenced by the styles and techniques it has learned.

**********

And how different really is that from learning from your teacher? Teachers, as humans, are themselves patterned. And patterns are transmissible. Children grow up to be adults that display patterns of behavior and thinking learned from their parents. A teacher or parent will teach something based on their own patterns (or schemata, as Piaget might put it). Imitation, mimicking, copying, adapting, collecting, morphing …

Because AI is essentially employing existing artistic styles and techniques it learns, it could be seen as a form of derivative work or imitation rather than original creation. On the flip side, you can argue that since the AI is generating new artworks based on learned patterns and not directly copying specific paintings, it does not meet the traditional definition of plagiarism.

Creating new works with existing knowledge is also how science works, building on the knowledge and techniques of those who came before. Creativity also comprises what we can do within certain bounds or with existing materials. It doesn’t have to be invention of something that never existed before. It can be evolutionary, which is the common way creativity manifests.

Other side of plagiarism

While the author ties herself too closely to copying original works, AI, and creativity, I would argue that there is a point to be made about plagiarism simply defined as “representing as your own work you did not do.” If someone takes language-model-produced content, which contains ideas and their expression the human author did not produce, that could be seen as plagiarism.

https://talkingonthethink.wordpress.com/2023/05/19/thoughts-on-ai-plagiarism/#more-2179