Raw LLM Responses

I'm a sci-fi writer, I think, but no matter how hard I prompt it and how many times I try, using the top-of-the-top premium models costing an arm and a leg, it sucks ballz. I am serious, I do better even with my low skills. Sure, it might be able to write the dialogues of this video, and some articles, a poem or a song, but once you try anything deeper it just cannot for a simple reason. It’s training. All LLMs go through training on large datasets. The problem is that they get something called overfitting and hallucinations, which cause them to go off script, forget important parts, or just rewrite the same story over and over again in very short paragraphs. Let's face it, to train an AI to be a good writer, one needs to be a good writer himself, and I doubt the geeks out there have that type of skill, and I doubt the new AI-to-AI training will help too. So sure many company leaders have already fired people in that hype that it will get better and replace humans, but it will still suck and at some point a firm that kept its talented staff will always do better. In clear, we come to a bottleneck that can only be resolved through specialization, and lots of funds or money, who will pay the bill? And speaking about money, let's cover that too. These models need a massive amount of power, energy, and computing resources to function, and who will be paying for that, your boss, check out his future bill when AI firms actually charge them for the actual cost of running these beast, he is gonna call you the next day to see if you still want your job back. You see for now the hype is so high AI company's to stay on top the curve must run at losses, but it is unsustainable and if it is not properly managed they will go bust. The reality is that these AI mega corps are for now footing the bill, and why? Because they expect profit. They aren't building Skynet for fun, you know they need a return ASAP, and for now none of them managed to be sustainable yet, it's a big bubble of investment ready to burst. Like with internet bubble, AI generative tools will crumble and with that only a few companies will be controlling the market, and it will make them susceptible to regulations and taxes etc. If they step out of bond they will be heavily sectioned and all investment to improve models will be subjected to tight inspection as with any product they are making. Politically I doubt people will vote for loosing more jobs to AI and bots so we will soon have strong policies stooping that growth, it is already happening in the EU. Now, coming to robots taking over and replacing us, I have just one other stupid question... With what power and resources? Because unless we give all our freedom and lives to mega-corps that design these machines and maintain them, there won't be enough energy and materials to help this expansion grow enough to take over. Another huge bottleneck. My vision of the future is that we will have some form of AGI and even something close to super intelligence, we will have robot clerks, street cleaner etc, but it will be bottlenecked by plain economic forces just like we are. I mean, imagine what anyone here can do with unlimited power and resources, huh? It's the same with AI. No power, No resources No great human replacement plan! Who would have known that the most deadly foe to AI global expansion is capitalism, politics and market forces? Enjoy!!!