Stochastic Parrots Day: March 17, 2023

Since the publication of On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?🦜nearly three years ago, many of the harms the paper has warned about and more, have unfortunately occurred. From exploited workers filtering hateful content, to an engineer claiming that chatbots are sentient, the harms are only accelerating. The listed co-authors of the paper and various guests reflected on what happened in the world of LLMs, in the two years since the paper's publication.

Screenshot of the linked image on sound cloud. There's "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big"? Playing and at the bottom is Emily M. Bender's headshot. On the right hand side is a sticker with a green parrot saying "Consent" and with the title of the paper writing in a cirlce around the parrot.

Audio Paper

Do you prefer an audio version of the paper? You're in luck because our paper has now been released in audio format. Take a listen.

Listen here
A screenshot of the linked website. Says " « On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜 »  Je vous propose un fil lecture et résumé de cet article de @emilymbender, @timnitGebru, @mcmillan_majora et @mmitchell_ai, qui expose avec clarté les enjeux actuels autour de l'IA appliquée au langage. Les autrices s'intéressent en particulier aux enjeux liés aux modèles de langage. Il s'agit de modèles d'IA dont la fonction est de produire des phrases, du texte.  Il peut s'agir de traduction, de sous-titrage automatique, de légender des photographies, etc... "

French Translation

Here is a summary of our paper translated into Français/French, by @CortexNihilo.

Read here

Stochastic Parrots Day Reading List

The 3,300 hundred attendees and us apparently discussed 137 works! Thanks to Mirabelle Jones, this reading list is now assembled in a doc that you can read.

Read here
A screenshot of the illustration and title from the Science for the People Article linked. The title says "Stochastic Parrots: How Natural Language Processing Research Has Gotten Too Big for Our Own Good." The illustration is mostly in purple, showing a woman touching a tree which looks like its made of circuitry.

Science for the People's Article

Esther Sánchez García and Michael Gasser wrote an article in Spanish for Science for the People summarizing our paper.

Read here
A screenshot of a white hoodie with the writing "Stochastic Parrot is Perplexed" on the hoodie, and a parrot next to a dice, and an equation that is not legible.

Stochasitic Parrot Hoodie?

Mark Riedl, who also was a panelist in our Stochastic Parrots Day events, designed a Stochastic Parrot hoodie, with proceeds donated to the ACLU.

Get yours here
A sticher with a green parrot saying "Consent" and a black circle around it with the text "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big" written in black

Stochasitic Parrot Sticker?

Speaking of merch, Ruth Starkman designed Stochastic Parrots stickers. She might have some left!

Ask her on twitter

Stephen Fry Talks about Stochastic Parrots!

We've gone mainstream. Stephen Fry talks about Stochastic Parrots, and even mentions Emily by name!

Watch here
Screenshot of the illustration and title of the linked article. On the left is text saying "You are Not a Parrot. And a chatbot is not a human. And a linguist named Emily M. Bender is very worreid what will happen when we forget this." On the riht is Emily M. Bender sitting on a chair, looking at a large red anad green parrot on her shoulder.

You Are Not a Parrot

While writing this thorough profile of Emily for New York Magazine, Elizabeth Weil reminds us that we are not parrots!

Read here
Screenshot of the title and illustration of hte linked article. Title says "Stochastic Parrot": A name for AI That Sounds a Bit Less Intelligent. "An ancient Greek word for guesswork fuels a term that suggests supersmart computer programs are just mimicking whatever they see" by  Ben Zimmer. With the illustration showing brown skinned woman typing in front of a screen wiht a green rectangular face on the screen, and in front of her somoen wearing what looks like a parrot costume, doing the same.

Stochastic Parrots is AI Related Word of 20223!

The American Dialect Society named "Stochastic Parrots" AI related word of the year. Read this Wall Street Journal article by Ben Zimmer to learn more about the phrase Stochastic Parrots.

Read here
Screenshot of the linked article, with title saying "Opinion: We warned Googel that people might believe AI was sentient. Now it's happening." And a picture of a person through glass.

We Warned About "Sentient AI"

Timnit and Margaret wrote an op-ed for the Washington Post reminding everyone that they had warned Google that people could believe its LLMs were sentient.

Read here
Screenshot of the linked article showing selected media coverage of stochastic parrots. Bender, Gebru et al 2021 (On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜, presented at FAccT 2021, also available as an audiopaper), has been broadly covered in the media. This page houses a collection of links to such stories. I am also collecting translations and translated summaries of the paper into various languages. Selected Media Coverage of Stochastic Parrots  We read the paper that forced Timnit Gebru out of Google. Here’s what it says., Karen Hao, MIT Technology Review, December 4, 2020.  AI ethics pioneer’s exit from Google involved research into risks and inequality in large language models, Khari Johnson, Venture Beat, December 3, 2020.  "I started crying": Inside Timnit Gebru’s last days at Google---and what happens next, Karen Hao, MIT Technology Review, December 14, 2020.  Inside a Hot-Button Research Paper: Dr. Emily M. Bender Talks Large Language Models and the Future of AI Ethics, Hayden Field, Morning Brew, February 1, 2021.  Wie viel Ethik verträgt Google?, Eva Wolfangel, Zeit Online, February 2, 2021.  AI and the List of Dirty, Naughty, Obscene, and Otherwise Bad Words, Tom Simonite, WIRED, February 4, 2021.  Kunstig intelligens-forsker: — På en måte er hele fagfeltet en skam, Hanne Østli Jakobsen, Morgenbladet, February 12, 2021.

More Media Coverage

Emily has curated a list of additional media coverage on stochastic parrots.

Read here

Sign up for our monthly newsletter