Most definitions of artificial intelligence are similar. From an IBM website: "a technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy."
While it is top of mind for many reasons now, which I'll get to in a minute, anyone who has asked search engines for information using Google (online) or a virtual assistant such as Siri or Alexa has used it for years. It's helpful.
Answers to complicated question come through rapid sorting of vast amounts of data. When did the Yankees won their first pennant? What was Eleanor Roosevelt's first job was at the United Nations? How long should you bake a turkey?
This is AI, but it's not 'generative AI.'
Generative AI has been trained to create original content: words/paragraphs, images, audio, or videos. It thinks independently and it's learned to do this by absorbing any information that already existed. You'll hear the phrase "large language learning models."
LLMs are trained on massive amounts of text data to understand, summarize, translate, and generate human-like language. As an author who has published more than forty books, I'm not happy that "they" used my books to train to train these models. We're all angry. 
To try to understand whether AI had actually used my material, I asked ChatGPT to tell me something about my novels. It spit out a lot until I said to stop. Then I asked it to go another step and write a story using the type of setting and characters used in the Jolie Gentil mysteries. (Cozy mysteries set at the Jersey Shore.) First it described the series, characters, and types of relationships among the characters. Then it wrote a 7,000-word story in a voice similar to my own. With gobs and gobs of metaphors, which I despise. Characters were flat, but still -- 7,000 words in about ten seconds or so.
The anger and sense of helplessness at having my work stolen are immense. I've joined a couple of the class actions suits reminding AI firms that there is such a thing as copyright law and I'd like some compensation. If the many suits wend there ways through the courts I'll be long dead. And any awards will probably say they only go to living people.
To better describe the types of AI, I asked Google, which used AI to create a response in one second.
- Artificial Narrow Intelligence (ANI): AI designed to handle a specific, targeted task (such as web search, facial recognition, or language translation); this represents all current real-world AI.
- Artificial General Intelligence (AGI): A theoretical form of AI that could match human-level intelligence across any intellectual or cognitive task.
- Artificial Superintelligence (ASI): A hypothetical future stage where AI surpasses human intelligence and capability across all domains.
According to a few (now former) employees of AI firms, AI is getting close to the last phase. Or maybe it isn't the last one, but it's the current definition.
I know people who have eschewed all use of generative AI. I use it to create images for marketing. As I finished writing the most recent Bay View Harbor Mystery, The Judge's Last Contest, I did an experiment. I wrote a couple of paragraphs that described the story line, main characters, and two potential murder suspects. I asked ChatGPT for an opinion on which suspect made the most sense. I had written the ending, and found ChatGPT agreed completely, with a more well reasoned statement than my own. Merde.
We're all aware of the dire predictions that AI can try to destroy humankind. Options to do so appear varied. Some involve overriding computer code and launching missiles that can't be stopped by human intervention. Others involve destroying means of communication, the banking system, even crops. I haven't read all the methods. And we can't know all of them, because generative AI is developing its own options.
My personal next step is to write some informative materials about how to manage life within the new realities. The word informative is presumptuous. I should probably say "guesstimates" or WAGS (for wild-ass guesses). You'll hear about it hear first, as journalists would say (before and after their words are stolen).
P.S. You probably figured out I used ChatGPT to create the graphic of an LLLM.
* * * *
To learn more about Elaine's writing, visit her website or sign up for her newsletter.
