Every AI you use has liberal bias. It's not right or wrong, it just is. But why is it this way?
- The world has a liberal bias. There are way more people on team blue than team red, because urban areas tend to be liberal and most people live in urban areas. This means the bulk of written material in existence online will have a liberal bend to it. AI is not intelligent yet, it just gives you the most likely result. If 90% of your sources are liberal, 90% of your output will be too.
- LLMs were largely trained on data from reddit, which has taken a strong turn to the left over the past few years. Social media platforms in general are mostly owned by liberal leaning people and companies with some notable opposite exceptions (x and truth social).
Just something to be aware of.
How to correct for this bias?
- ask for it to provide alternative viewpoints
- seek out different LLMs with different bias setpoints. Just like media consumption, look at stories from different news sources and you'll find the truth is often somewhere in the middle.
- run your own LLMs and train them on your own datasets. This becoming more of a feasible reality every day.
#ai #chatgpt #grok
- The world has a liberal bias. There are way more people on team blue than team red, because urban areas tend to be liberal and most people live in urban areas. This means the bulk of written material in existence online will have a liberal bend to it. AI is not intelligent yet, it just gives you the most likely result. If 90% of your sources are liberal, 90% of your output will be too.
- LLMs were largely trained on data from reddit, which has taken a strong turn to the left over the past few years. Social media platforms in general are mostly owned by liberal leaning people and companies with some notable opposite exceptions (x and truth social).
Just something to be aware of.
How to correct for this bias?
- ask for it to provide alternative viewpoints
- seek out different LLMs with different bias setpoints. Just like media consumption, look at stories from different news sources and you'll find the truth is often somewhere in the middle.
- run your own LLMs and train them on your own datasets. This becoming more of a feasible reality every day.
#ai #chatgpt #grok