Data Ethics Club meeting 17-02-21, 1pm-2pm GMT #

Meeting info#

Zoom link: https://bristol-ac-uk.zoom.us/j/99654848727?pwd=dzUvQUErTGREYUNzb3ExMnJON3pDUT09

Link to content: http://faculty.washington.edu/ebender/papers/Stochastic_Parrots.pdf

You’re welcome to join us for our next Data Ethics Club meeting on Wednesday 17-02-21 1pm-2pm. You don’t need to register, just pop in. This time we’re going to read On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? by Emily M. Bender and Timnit Gebru (joint first authors) and colleagues, which is the paper that got Timnit fired from Google.

Thank you to Nina for suggesting this week’s content, and to Nat, who’ll be leading this week’s meeting.

There will be time to talk about whatever we like, relating to the paper, but here are some specific questions to think about while you’re reading.

Do you find it (counter-)intuitive that AI-generated text is without meaning, i.e. that the AI is a “stochastic parrot”?
What are the implications of the meaninglessness of AI-generated content in other contexts?
What are the implications of the fact that Google did not want this paper published?

Other things to think about:

At the beginning of the meeting, we’ll start by inviting an attendee to volunteer to summarise the paper: we would welcome a ~3 minute summary of the paper.
This paper is more technical than our previous ones. In any good journal club, the discussion is an opportunity to ask questions about any bits that you didn’t understand, so keeping a note of any questions might be helpful in sparking discussion, too.

Number of people: 16

14/14 voters agreed that the content sparked interesting discussion. 14/14 voters would recommend the content to others.

Nat did a great summary of our discussion on Twitter. You can find it here!