Ethical and social risks of harm from Language Models

What is an example of exclusionary norms in language models?

What type of harm can occur when language models reinforce stereotypes?

What risk does the automation of tasks by language models pose to job quality?

The environmental impact of operating language models is less significant than the impact of training them.

A LM that is only correct 50% of the time is arguably more dangerous than one that is correct 99% of the time.

What type of information can language models inadvertently reveal?

LMs have been used to influence government elections.

How can inference capabilities of language models pose a risk?

How might language models contribute to income inequality?

What is one of the primary causes of toxic language in language models?

Misinformation is only a serious concern in sensitive domains

Language models may lead to increased job quality by reducing monotonous tasks and allowing for more creative work.

What mitigation approach can be used to reduce privacy risks in language models?

Why AI-driven surveillance and censorship problematic? (Select all that apply)

How can language models compromise user privacy?

In what way does common token bias affect LM misinformation?

What is a “common token bias” in language models?

Humans are unlikely to trust conversational agents to the point of blindy following dangerous or illegal advice.

What challenge is presented by the access to language models across different populations?

How can language models promote unfair discrimination?