Stargazer is a simulated intelligence system - basically a chatbot. I’m trying to get it to feel very close to AI though. With just over 600 blocks, Stargazer alters pre-written phrases by find suitable emojis, synonyminize words and customise the response to match input context. Plus, responses are randomised yet again by translating it through either French or Danish, then back into the system language. It also has an awareness of context - a couple of tokens from the last input is added to the next one. Some grammar mistakes can still sneak in, but the translator will squash out as many as it can. Responses are chosen through testing if each word of the input matches or conflicts with a given response. This is done through an algorithm - thankfully not a giant ‘if’ statement - and the keywords and responses are in a list. It’s very easy to add new information to the database. Then, words are swapped out with similar words, and translated through different languages to add a small bit of variation. Commands can be used too. /add - Adds some info to the database. /lang [language] - Change the output language. Note that inputs must always be written in English. /getcontext - Get the context. /clearcontext - Clear the context.
Currently only a few words are available, because I’m trying to get the system perfect before adding new data. Please report anomalies or strange/unrelated responses, I’ll try to fix them and release a new update soon. There are a few pre-written Easter eggs, like phrases from 2001 Space Odessy or IHNMYIMS. You should be patient with it! Sometimes it can take a while to find the best response, especially if your question is ambiguous or long. UPDATE LOG: Apr 29 2026 - Added the context system. Jan 3 2026 - Found errors with the scoring algorithm, fixed. Responses are now more related to the input. #ai #chatbot
