Rozado's research uncovered significant political biases within LLMs, with a prevalent left-of-center stance, raising concerns about AI’s influence on public attitudes toward various social issues.
Most existing LLMs display left-of-center political preferences when evaluated with a variety of political orientation tests, highlighting the biases that exist in the technology.
The custom bot tests showed that AIs could be influenced to express political leanings based on the training data provided, illustrating the malleability of LLM biases.
The worry is that our thinking could be affected by LLM responses, as these models begin to displace traditional information sources like search engines and Wikipedia.
[
Collection
]
[
|
...
]