How ChatGPT mangled the language of heaven | Letter

Asked to generate a story from an English translation of a letter in Welsh published in the Guardian, the AI chatbot came up with a lot of twaddle, reports Fiona Collins

Ian Watson (Letters, 17 February) asks for a translation of my letter in Welsh (13 February). I did include an English translation in my letter, but only the Welsh was published. I sent a second letter asking the Guardian to publish the translation, as I was having a lot of stick from a certain friend who couldn’t read it, but with no luck. Hopefully Ian’s letter will change the letters editor’s mind.

The English version was as follows: “Thank you very much for the excellent editorial article which sang the praises of the Welsh language … Since you are now so enthusiastic about Welsh, may I, from now on, write to you in the language of heaven?”

Meanwhile, there has been much glee about my letter on Welsh-language social media. Furthermore, a storyteller friend who doesn’t speak Welsh fed it into Google Translate, and got a pretty accurate English version. He then fed the translation to ChatGPT, the artificial intelligence chatbot, and asked it to construct a story based on the letter.

Alarmingly, but unsurprisingly, the chatbot produced a lot of twaddle in which the Guardian editor and I fell in love, as a result of our shared passion for the “language of heaven”, and lived happily ever after. I don’t think ChatGPT realised that iaith yr nefoedd (language of heaven) is a term used to describe Welsh. Though whether anyone has authenticated if it is spoken there, I sadly can’t tell you.
Fiona Collins
Carrog, Sir Ddinbych

The GuardianTramp

Related Content

Article image
Ignorance is not bliss for ChatGPT | Letters
Letters: Geoff Renshaw has realised that the chatbot can’t say ‘I don’t know’, Tom Brown thinks it has been reading too much Kafka

10, Sep, 2023 @5:33 PM

Article image
The Guardian’s block on ChatGPT using its content is bad news | Letters
Letters: Praveen Naidoo says there is need for a wider debate; Paul Sawbridge and Ian Watson believe the information fed to AI should not just come from the rightwing press

04, Sep, 2023 @5:02 PM

Article image
The Y2K bug should teach us to be wary of AI | Letters
Letters: John Thow on the global effort that prevented a Y2K disaster, and Robert Frazer on Isaac Asimov’s laws of robotics

08, Jun, 2023 @4:58 PM

Article image
The danger of blindly embracing the rise of AI | Letters
Letters: Readers express their hopes, and fears, about recent developments in artificial intelligence chatbots

03, Apr, 2023 @4:52 PM

Article image
Press pause on the use of artificial intelligence in public services | Letter
Letter: AI’s impact on social cohesion and sustainability must be addressed, writes Dr Melanie Smallman

01, Nov, 2023 @5:35 PM

Article image
The revolutionising potential of medical AI | Letters
Letters: Readers respond to news that Theresa May has promised millions towards artificial intelligence that could help fight cancer and other diseases


22, May, 2018 @5:02 PM

Article image
Human error, not artificial intelligence, poses the greatest threat | Letter
Letter: The risk that humanity faces comes not from malevolent machines but from incompetent programmers, writes Martyn Thomas


03, Apr, 2019 @5:28 PM

Article image
We need a much more intelligent approach to the rise of AI | Letters
Letters: Readers respond to Larry Elliott’s article about the impact of artificial intelligence on the workplace

14, Apr, 2023 @4:48 PM

Article image
The existential threat from AI – and from humans misusing it | Letters
Letters: Roger Haines writes that there is no evidence of AI sentience, while Prof Paul Huxley recalls Asimov’s laws of robotics, and Phyl Hyde says fears of AI are being overblown

02, Jun, 2023 @4:16 PM

Article image
The benefits of AI and machine learning | Letters
Letters: Prof Rose Luckin, Anthony Seldon and Priya Lakhani say artificial intelligence is not to be feared and point out how it can help students


30, Sep, 2019 @5:00 PM