Episode 27: Language awareness in the age of AI


In the introduction, we mention a number of  language-related AI applications, including DALL-E, which generates images from language prompts; Scite, which identifies references supporting or questioning research findings; ELSA, which stands for English Language Speech assistant and is meant to help language learners; Wordtune, which can rewrite texts i different “tones”; and the codings apps Copilot and CodeWhisperer, which convert language inputs into code.

Some examples of how the AI-powered version of the Bing search engine produces answers that are troubling or face-threatening are mentioned in this article.

AI (Dall-e) generated image

Our interview guest, Andreas van Cranenburgh, refers to how OpenAI, the company behind ChatGPT, uses low-paid content moderators in developing countries, often exposing them to traumatic content. This practice was described in Time Magazine.  

Following the interview, we talk about how the notion of communicative competence needs to be extended for interactions with chatbots.  Dell Hymes published his seminal paper back in the early seventies: 

  • Hymes, D. (1972). On communicative competence. In I. B. Pride & J. Holmes (eds), Sociolinguistics (pp. 269-293). Penguin.

He revisited the idea twenty years later:

  • Hymes, D. (1992). The concept of communicative competence revisited. In M. Pütz (ed.) Thirty Years of Linguistic Evolution (pp. 31-57). Benjamins.

Hymes’ original formulation of communicative competence dates from six years after the first ever chatbot, Eliza, was developed. (It is not known if he was aware of it.) The creator of that application, Joseph Weizenbaum, named it after Eliza Doolittle, the character in Bernard Shaw’s play Pygmalion (later made into the musical and film My Fair Lady). In that modern take on the Greek myths of the sculptor Pygmalion, who falls in love with one of his statues, a linguistics professor teaches a working-class woman how to sound upper-class. Are chatbots the malleable female creations of male developers? And why does Erika, a female user, think of ChatGPT as a man?

As they say, there is a paper in that.

And finally, here is the ChatGPT-generated text we analyse in the last part of the episode:

Dear [Customer Name],

Thank you for reaching out to us. We understand that high energy prices can be frustrating and we want to help. We’re sorry for any inconvenience this may have caused you.

Our company’s energy prices are affected by a number of factors, including changes in the global energy market and increasing demand for energy. However, we are committed to finding ways to help our customers manage their energy costs.

We recommend some simple steps to conserve energy, such as turning off lights when they’re not in use, adjusting your thermostat, and using energy-efficient appliances. Additionally, we offer a number of energy-saving programs that could help you save money on your energy bills.

We value your feedback and appreciate your loyalty. If you have any further concerns or questions, please do not hesitate to contact us.

Best regards,

[Your name] 

Our next episode will conclude the mini-series in CSR – see you then!

Listen to the episode here

Full transcription of the episode


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s