AOW: Chatbots

What is one of the problems associated with AI systems like ChatGPT as described in the article?

• They require too much computational power.

• They sometimes make things up, described as hallucination.

• They are unable to understand human language.

• They are too expensive for most organizations to use.

According to the article, what is the potential risk of using AI on tasks with high-stakes consequences?

• AI systems may not be able to handle large amounts of data.

• AI systems may produce false information.

• AI systems may be too slow in processing information.

• AI systems may be too difficult for the average person to use.

Who is Daniela Amodei and what is her stance on AI hallucination?

• A linguistics professor who believes AI hallucination is fixable.

• A co-founder and president of Anthropic, who states that all models suffer from some hallucination.

• A tech executive who is optimistic about overcoming AI hallucination.

• A high school student who uses AI to write legal briefs.

What is the estimated value of the generative AI technology to the global economy according to the McKinsey Global Institute?

• $2.6 billion

• $2.6 trillion

• $4.4 billion

• $4.4 trillion

What is Emily Bender's position on the fixability of AI's hallucination problem?

• She believes it is an inherent issue that cannot be completely fixed.

• She is confident that it will be fixed within a year and a half.

• She thinks it is not a serious problem and can be ignored.

• She claims that new AI models have already solved the problem.

What does Sam Altman predict about the future of AI's hallucination problem?

• It will be completely solved with no trace left.

• It will be much better but not perfect, and a balance between creativity and accuracy is needed.

• It will become a more severe problem over time.

• It will not affect the performance of AI systems in any significant way.

What is the main function of language models as described by Bender?

• To create new forms of entertainment

• To model the likelihood of different strings of word forms

• To generate entirely new languages

• To replace human writing in all aspects

Which of the following is NOT a feature of the latest crop of chatbots like ChatGPT, Claude 2, or Google’s Bard?

• They can generate new passages of text

• They are designed to make things up

• They can always select the most plausible next word in a string

• They have an "autocomplete" feature for composing texts

According to the article, what is a concern when language models are tuned to be right more of the time?

• They become less creative

• They are harder for a person reading the text to notice

• They require more computational power

• They become more expensive to use

What does Shane Orlick say about the customers' feedback regarding Jasper AI?

• Customers are concerned about the security of their proprietary data

• Customers are unhappy with the hallucinations of the AI

• Customers tell how Jasper AI has helped them with ideas they would not have thought of themselves

• Customers believe Jasper AI's answers are the most trustworthy