Most people over the age of 50 recall a time when the news was delivered expressionless and deadpan. The job of the journalist was to present facts and nothing else. Any hint by a journalist of agreement, disagreement, Schadenfreude, favoratism, etc often resulted in the journalist being fired and it could have just been a smile that caused the firing. We find that now-a-days, regular news articles read more like Op/Eds and steer the reader in a specific direction or use more favorabke language around something the author agrees with and distain for things the author disagrees with, even villifying people who don't deserve villification... instead of just presenting facts in an unbiased manner... and nothing else, no leading language. The author of the article should let the reader make sense of the facts in the article based on the reader's own life experience. For this reason, to maintain completely fair objectivity, the Nonpartisan Report takes news articles from various sources, extracts the facts, points out the bias. To be completely fair, the analysis is generated by AI. Although the views, opinions, and biases of the AI itself may not necessarily be shared by the Nonpartisan Report humans, we think it does a fairly decent job some of the time. Please keep in mind, AI has made some huge gains in recent years, but it still kind of sucks at some things. It is not the intention of the Nonpartisan Report to offend anyone, but the AI may offer some analysis on polarizing topics that many people may consider offensive. If you find yourself offended by the AI analysis, please understand that offending you is not our intention and treat the AI output as if it was generated by an articulate 4 year old.
Note, we are currently running the AI locally on a moderately equipped consumer grade PC that does not have a fancy video card - and it is most likely a quantized version of an LLM... so it is going to fumble from time to time... and it takes so much time to generate the analysis. Also note, we don't plan to moderate the content here. If it becomes a problem we will, but we are hoping to remain as hands off as possible. As new language models come out, we'll probably change the LLM (or whatever ends up replacing LLMs) being used for the Analysis, and we may also change the prompts from time to time as well, get new hardware, etc... and hopefully the content produced will get better and better with time. But, as of May 2024, it is "OK" - so take everything with a grain of salt and please be patient with us as we experiment with the technology and also be forgiving of the LLM's transgressions and again, also be forgiving of us humans. Over time, we will create analysis for more news sources.
Technical note: we will specify which model we used at the end of each AI anlysis - just assume we are using the smallest version of a family of models most likely running on Ollama although we may experiment with various API servers from time to time as well, and yes, we have found using different API servers does impact the output.
The opinions of the AI are baked into the language model during training, possibly sometimes unintentionally by the LLM's creators. We did not have any part in the LLM training, wieights, etc. although we do control the prompts which are a work in progress and proprietary. The satire call out is a political statement that most news meadia outlets now-a-days are extremely biased to the point of further driving a wedge between ideologies and have taken a nation of friends and neighbors who, in the past, might have a mild disagreement in ideologies, now are at a dangerous precipice where 1/2 of a nation hates and fears the rest of the nation. We feel that this been intentionally for ad revenue and agenda pushing.
This site, and it's content, should legally be considered political satire and the apparent non-anthropomorphized net resultant opinion of an arbitrary AI language model.