Does AI dream of electric sheep?
A few months ago it was just a background whisper, while now it seems indispensable in our lives. I’m talking about AI, artificial intelligence that has replaced the most disparate operations.
Do I need a recipe for apple pie? I ask AI, which is faster than Google. Do I need to quickly translate an entire article into multiple languages? I ask AI, which makes fewer mistakes than Google. Do I need to waste some of my time? I ask AI, which is more fun than Google.
In short, this Intelligence seems better than the boomer parent in every way. Not for nothing has it earned the reputation of internet within the internet. A bit like Her, but without the voice of Scarlett Johansson.
Many have seen it as a revolution, but others fear the personal offense. “What if it takes our jobs? What if our children stop thinking? What if the world becomes the steel abyss predicted by Asimov?”
Ethically, the issue is complex, but thinking that contemporary AI can really replace humans is premature for many reasons.
The entropy of the text
First of all, AI bases its Intelligence precisely on us. If the internet became a network populated only by content produced by AI, the AI itself would not know what to do.
Several field tests have shown that all the best AI detector tools can easily distinguish between texts written by artificial intelligence and human-written ones. In fact, even though at first glance they may seem identical, the former are easily recognizable for their predictability. In essence, AI works according to what in linguistics is called expectancy grammar or the grammar of anticipation, a series of probabilistic mechanisms according to which, based on the context, one word is preferred over another. Every day, each of us does this unconsciously, for example, when we listen to a sentence, mentally anticipating what will come next.
However, even though our words have a certain degree of predictability by their nature, the variations that our brains decide to apply are too numerous for a machine (such as an AI detector) to really predict the final form that the sentence will take. On the other hand, this is different for texts generated by AI, which are always statistically predictable.
The will to create
The question only arises when it comes to the creative field of the human being. I doubt anyone feels threatened by their washing machine just because it washes clothes better. It is designed to do so, otherwise, it would not be useful in any way. The real difference between humans and AI, therefore, lies not in the end result, but in the will to create something. In fact, as limited as our abilities may be compared to those of AI, they will always start from a desire to be expressed. Conversely, AI waits for a command to produce something and never takes the initiative. At least for now.
Writing a book with AI
In this sense, artificial intelligence could be an excellent tool for those who have ideas but lack the skills to put them into practice. Already, it seems that many books written or co-written with ChatGPT can be found on Amazon, mostly on the topic of AI, but also collections of poetry and children’s books. In practice, one defines their own stylistic uniqueness on a product that has already been semi-finished by artificial intelligence.
An example of this use is given by merchant Brett Schickler. In an interview with Reuters journalist Greg Bensinger, he explained how AI finally gave him the opportunity to write a children’s book. Starting from OpenAI’s chatbot, Schickler created a 30-page text with illustrations (also generated by artificial intelligence), which he then sold through Amazon Direct Publishing in both ebook and paperback formats. Although the earnings were not very high, the merchant said he was satisfied and ready to do it again because AI filled his literary and graphic shortcomings and saved him time in developing the project.
Intellectual property
The issue of intellectual property still needs to be addressed, as it remains unresolved. Amazon has stated that it does not intend to modify its publication rules following the use of AI, as long as self-published texts do not contain parts for which the copyright holder does not have the right.
The problem is that OpenAI’s chatbot, currently the most used for this type of publication, does not seem very comfortable with the question. In OpenAI’s terms of use, it appears that both the input and output are owned by the user: “You can use Content for any purpose, including commercial purposes such as sale or publication, if you comply with these Terms.” However, in response to a direct request, the chatbot says: “If you use my texts for commercial purposes, such as selling them to your customers or monetizing them on your website, you are violating OpenAI’s intellectual property rights. To use my texts for commercial purposes, you need OpenAI’s explicit permission.”
For some, it may seem like a deception, while for others, it is a simple way to make their dreams come true. For now, we can be sure that the realm of AI is not yet populated by electric sheep.
