Raw LLM Responses

AI is a mirror. I've been using and testing them for a year now. The mindset and competence of the human determines how well the AI works. I know this and have proven this. I am a okay programmer, not a professional either. In just two weeks, I created a AI driven NPC framework in Opensimulator that works with both remote and local AI. The AI/NPC has pathfinding with predicitive targeting to follow and keep up with a player. It is spactually around and can talk about everything around it with minimum hallucinating and when I say minium I mean it doesn't say things are there that aren't, but embleshes and fills in the blanks about the descriptions of what is around them. It can hold mulitple conversations with multiple players while staying in character. The AI is able to control multiple NPCs each with their own appearance and personalities. It can make decisions about where to seat based upon both a blacklist and scoring system. It is just 6 scripts and 3 configuriation notecards so that non-programmers can customize each npc in dozens of ways without touching the code. Each script's timing is staggard to keep from overloading the server across mutliple NPCs. Also the AI/NPC has facial and physical gestures that it can express based upon what is sad to it. I did all of this from concept, design doc, pseudo code, translation, testing, to completion in just two weeks in a language I knew little to nothing about. Now, you may not be impressed by that, but I was never frustrated or overworked. I collaborated with it, in a iterated process. You had a bad experience, it was more about you and your mindset than it. Trust me, I've tested ChatGPT, Claude, Grok, Gemini, Deep Seek, Copilot for over a year, starting with small arcade games like Astroids, in Python, Lua, Java, Javascript, C, Common Lisp, Racket, Forth. They are better with Java, Python, Javascript, and C, and Java appears to the best language when it comes to its strong type, clear errors, and restrictive nature. Forth is a nightmare to LLMs, and so was Racket and CL, because they are very outside of bulk of their training, and those languages are difficult for most programmers to keep in thier head. Now is AI for everyone, nope. Just like any tool it depends on the skill and mindset of the user. If your skills are good, your good at communicating, critical thinking, and your mindset is compatable, it will be a performance multiplier, but it is also a multipler of the programmers traits. Lazy, vague, illogical and contradictive thinking equals bad code. Also coming into it with a negative perception and mindset about collaborating with the AI will just frustrate and make bad code. Now, I created that AI driven Framework with no special IDE either, just conversation, notepad, and the built in IDE in Opensimulator. I copied and pasted the old fashion way and still got it done in just two weeks. Its on Github, look it up, and there's an article about it on Hypergridbusiness online mag. Look it up, its opensource and free for non commerical use. AI is here and everywhere, its a tool but how good of one depends more on the user/company.