Your misunderstanding is shared by the entire machine learning industry and I've addressed it here multiple times regarding the use of lossless compression as the least biased model selection criterion possible. I'll try to present the 2-part argument, again, in a more concise manner:
1) If you have some reason to believe some particular set of data is "biased", then IF you are going to even *pretend* to participate in the social process called "science" you are obligated to share with the scientific community the *data* you believe supports your definition of "bias". 2) Having presented your data, it will then be appended to the set of all data being compressed into a macromodel of reality and contribute to its compression by bringing more dimensions of the data into consilience. e.g. If you think my thermometer is off by 1degree F, and you are correct, then your data in support of this will tend to bring measurements reported by my thermometer into the consilience with physical theory -- rather than challenging or "falsifying" physical theory. This 2-part process is to be repeated until all parties have exhausted their arguments over "bias in the data". On Sat, Jul 2, 2022 at 2:04 PM <[email protected]> wrote: > If the data to train an AI on says something lots like "women are usually > nurses", this it is so then, but the programmers want it to change, so, > they do have some weight in being the fathers of their AIs, unfortunately. > > Also, I heard these AIs can have a hard time seeing that males are more > often nurses even though the data shows it to be so. Or maybe if the data > lacks to show it. So, they want to correct that then. They call it bias. > Must rid the bias. > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T2f4774767b960b89-Mab0b781f8b0775aaed7c2627> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T2f4774767b960b89-M0c75f40c635a5dd6992fcfb0 Delivery options: https://agi.topicbox.com/groups/agi/subscription
