LiberLion on Nostr: The #AI And Human Bias. Many denounce the AI of intentional bias. Key concept: AI is ...
The #AI And Human Bias.
Many denounce the AI of intentional bias. Key concept: AI is plausible rather than true.
I will briefly explain how AI LLM Technology works, which is the most widely used today, then what an informational bias is and why it is a human outcome.
The future of Artificial Intelligence will be Super Synthetic Intelligence (SIS).
Let's see.
LLM models structure their responses based on a predictive text model that seeks to be plausible but will not necessarily be true. It is key to understand plausible as opposed to true.
The prediction thus coherently structures the idea based on words that follow others more likely, and that make sense with all the “learned” (trained) information.
Key concept: text prediction structure.
Thus, LLM technology is created with human information, not just the compilation but the predictive programming.
How do you expect it to be free of human bias?
What is an information bias?
An information bias occurs in data collection, analysis, or interpretation, leading to an erroneous or distorted conclusion. This bias may be intentional or unintentional.
The best known are sample selection bias, confirmation bias, anchoring bias, omission bias, and a very long etc.
So what will be the likely solution to minimize or avoid bias?
Let artificial intelligence generate a scientific method for information gathering, verification, and reproduction moreover, the AI generates its synthetic information.
There we will find ourselves in the era of the Singularity, the Synthetic Superintelligence (SIS).
Many denounce the AI of intentional bias. Key concept: AI is plausible rather than true.
I will briefly explain how AI LLM Technology works, which is the most widely used today, then what an informational bias is and why it is a human outcome.
The future of Artificial Intelligence will be Super Synthetic Intelligence (SIS).
Let's see.
LLM models structure their responses based on a predictive text model that seeks to be plausible but will not necessarily be true. It is key to understand plausible as opposed to true.
The prediction thus coherently structures the idea based on words that follow others more likely, and that make sense with all the “learned” (trained) information.
Key concept: text prediction structure.
Thus, LLM technology is created with human information, not just the compilation but the predictive programming.
How do you expect it to be free of human bias?
What is an information bias?
An information bias occurs in data collection, analysis, or interpretation, leading to an erroneous or distorted conclusion. This bias may be intentional or unintentional.
The best known are sample selection bias, confirmation bias, anchoring bias, omission bias, and a very long etc.
So what will be the likely solution to minimize or avoid bias?
Let artificial intelligence generate a scientific method for information gathering, verification, and reproduction moreover, the AI generates its synthetic information.
There we will find ourselves in the era of the Singularity, the Synthetic Superintelligence (SIS).