Ethics in Language Technology

What are chatbots commonly used for?

• Online customer service and support

• Playing video games

• Building websites

Why is bias in Natural Language Processing (NLP) tools and Large Language Models (LLMs) a concern?

• It slows down the AI

• It limits the AI to only one language

• It can lead to unfair or harmful outputs

What are word embeddings in language technology?

• The dataset used for training the AI

• A type of spell-checking tool

• Mathematical representations of words

What is meant by hallucinations in Large Language Models (LLMs)?

• 1. The AI seeing things that aren’t there

• 1. The AI generating information or answers that are incorrect or nonsensical

• 1. The AI translating languages inaccurately

What is an example of demographic bias in AI systems?

• Social media algorithms that show more content on a specific topic

• Facial recognition algorithms that are less accurate for people with darker skin tones

• Voice recognition software that struggles to understand certain dialects

What is a characteristic of a Large Language Model (LLM)?

• LLMs produce output based on a probable combination of sentences from their training data

• LLMs use predetermined rules to generate an output

• LLMs randomly select sentences from the data they are trained on to generate an output

What is NOT an example of a Natural Language Processing (NLP) tool?

• Spellchecker

• Paraphrasing tool

• Translation tool

• Voice assistant

What do you think of this presentation?

• Interesting! 😀

• It was okay. 🙂

• Boring! 🙄

• I don't know. 🤷‍♀️🤷‍♂️