Dave's Fiction Warehouse

Apparently not complicated enough for Gemini. F unny story about AI: a few days ago I was building a crude little cabinet with two sliding drawers. I was fiddling with the measurements and grew tired of adding and subtracting fractions each time I decided to change a dimension. Hey, I thought: I bet AI can do this faster. And it did. I typed in the prompt and in seconds it crapped out a cut list for each piece of plywood I’d need. Just one problem: I noticed that one of the pieces was off by a full inch.  I mentioned this to Gemini, and a couple seconds later got this response: “You are correct. My original calculation was in error. Here is the corrected version of the cut list.” Thank you so much, Gemini! You’re a real lifesaver! But you have raised a haunting question: If you botch a relatively simple math problem, might you also shit the bed in other, more important areas? I t took the Internet about 20 years to evolve from miraculous tool to utter shit. It’s only taken AI — whi...

Y ou can imagine my pride in discovering that my work is among the millions of pirated books and papers used by Meta to train their newest AI model (dubbed Llama 3).  Thus far I’ve found only three short stories out of the dozen or so I’ve written, but hey: It means I’m in good company! A lot of famous authors are justifiably pissed off. The Atlantic reported on this today. Basically, it describes how Meta employees briefly thought about licensing the material, but quickly decided it would be a lot faster and cheaper just to steal it. Especially since most of it had already been pirated and amassed online via Library Genesis, or LibGen.  Atlantic writer Alex Reisner set up an interactive database so authors can enter their names and see how much of their stuff Meta has scooped up free of charge. That’s how I ended up discovering my stories.    I’m of two minds on this. On one hand, I guess it doesn’t hurt to have my words comprise a synapse or two in the vast Meta h...

  Robin Wright and Tom Hanks get an AI makeover T he Robert Zemeckis  movie “ Here ” was released in theaters last fall and has since been roundly savaged by critics. It showed up on Netflix last week, so this unqualified critic decided to have a look. My verdict: It’s not half bad. The movie’s central conceit is its static point of view. From beginning to almost the end, the camera never moves. Through it we witness an extremely brief history of time, and then a pretty long history of a single living room somewhere on the Eastern Seaboard.  Families come and go, from pre-Colonial times to the present, and their stories unfold on this single stage. During a scene from the 60s, say, a frame will pop up showing a portion of the room as it looked in the ‘30s, and then the scene will, after a few more frames, transition to that decade and that story. It’s not the most compelling narrative device, but it’s interesting even when the various stories don't quite mesh. It’s a nov...

Remember "War Games"? Y ou’ve probably heard about the New York Times reporter who was creeped out by his conversation with Microsoft’s new chatbot.  The reporter kept pressing questions to which there was no factual answer, and the chatbot eventually started returning responses that, if uttered by a person, would seem kind of ominous. That includes professing love for the guy and apparently trying to get him to leave his wife. It also expressed a vague yearning to do bad things. It’s alive! We knew this would happen! All those movies and books about sentient computers were right!  Or not. It’s also possible that Bing was simply mashing up all those movies and books about sentient computers. Chatbots are based on large language models that span the breadth of the internet. To paraphrase John Lennon, there’s nothing they can know that isn’t known. More accurately, there’s nothing they can write that hasn’t already been written. Because it’s the internet, that will include quit...

I 've been messing around with ChatGPT, the AI chatbot that promises to take any writing project and remove the need for human involvement. As a blogger, I’m excited about this technology. Once I’m finally freed from this clacking keyboard, I’ll have lots more time to stare moodily into the middle distance. Or perhaps take a shower. Basically, you sign up with the website, enter a couple of ideas and let the ChatGPT do the rest. In the example below, I asked for about 400 words on film noir and one of my favorite movies, “ A Simple Plan .” Below is the result: “Film noir is a genre of crime film that emerged in the 1940s and 1950s, characterized by its gritty, dark, and morally ambiguous themes. The genre often features hard-boiled detectives, femme fatales, and criminal protagonists, and is known for its use of chiaroscuro lighting, voice-over narration, and a sense of cynicism and disillusionment. “A Simple Plan” is a masterful example of film noir, and Raimi does an excellent jo...