Tom Baudewijn Profile picture
Feb 14 13 tweets 7 min read
Last night in the #EPlenary @d_boeselager was probably the first MEP ever to use generative AI to write his speech. #1nt #CAI

For a look behind the scenes, a 🧵

europarl.europa.eu/plenary/en/vod…
As if a general AI speech was not enough of a challenge for the @EP_Interpreters, Mr Boeselager asked the AI to tell him how to regulate AI in Shakespearean English.
Humor and poetry, interpreters’ favourite snack, right?! 😅
Luckily, @EP_Interpreters and @Europarl_EN, can rely on a document and speech management system for meeting preparation and @d_boeselager was kind enough to upload his speech hours before delivery.
I was working in the NL booth when the speech popped up in our system and I started wondering how I would go about it. Instinctively, I thought: since it was written by AI, let’s ask AI to translate it for me.
In my first attempt, I asked #ChatGPT to retain the same style as the original. I was not convinced but I don’t blame the AI. There is no such thing as Shakespearean Dutch. ImageImage
My second idea was to ask #ChatGPT to #localize it, similar to what translators do for marketing texts or video game translations. This is what it produced in the styles of Gerard Reve (left) and Louis Couperus (right). ImageImage
Again, the result was not something I’d be comfortable with using in the booth. Too much is lost, in my opinion: the meaning is not clear at all nor does it rhyme as the original does.

By the time MEP Boeselager delivered his speech, my colleagues had taken over.
My Dutch colleagues decided not to interpret it and let listeners enjoy the AI’s attempt at Shakespearean poetry. Same for the Maltese, Polish and Slovene booths. All others did give it a go and it’s interesting to see how they tackled it.
I listened to the renditions in my passive languages. They all differ & it would be great to hear about my colleagues’ prep.

My German colleague rendered a beautiful version on rhyme and, breaking into a smile, you can hear her having fun. 😀

europarl.europa.eu/plenary/de/vod…
The Italian booth used a caveat right at the start by adding ‘The speaker is quoting Shakespeare.’

europarl.europa.eu/plenary/it/vod…
My French colleague interpreted the meaning and at the end added: ‘Unfortunately, the interpreter was unable to render the Shakespearean style in French and apologises for this.’

europarl.europa.eu/plenary/fr/vod…
Some comments:
-#ChatGPT had issues in the Safari browser, it worked in Firefox, though.
- It took #ChatGPT several minutes to translate it into the style of a Dutch author.
- I need to work on writing better #ChatGPT prompts.
To sum it all up.

Fun gimmick by @d_boeselager which could potentially fill an entire class on AI, the use of #ChatGPT for interpreters and kick off a discussion on simultaneous #1nt strategies on how to deal with humour and poetry in general.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Tom Baudewijn

Tom Baudewijn Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @DutchTerp

Oct 26, 2020
Since early September I have worked in different RSI settings. Some more polished than others. These RSI experiences may have changed my views on RSI compared to 👇 thread in March. Time for an update. #1nt #r1nt #rs1nt #remote #WFH
Let me first briefly share what types of RSI setups I have worked in. I’m sure many colleagues will be having different experiences. I’d love to hear them and read your views on how you see this evolve.
1/ Hybrid meeting at the @Europarl_EN : 1 interpreter per booth, some delegates in meeting room, others connected from home/office via @InteractioApp.
Read 24 tweets
Mar 13, 2020
I really have a problem with this narrative that interpreters have been reluctant about #r1nt because of ideology. It sounds very dismissive of interpreters’ concerns regarding #r1nt. In my eyes those concerns are very legitimate and much more than ideology. A thread. 1/
I want to make clear from the start that I see a big distinction between #r1nt from home versus from a hub. There are many serious issues with the former while I see opportunities and some issues with the latter. /2
What we see now due to #COVID19 is #terps considering #r1nt from home and giving it a chance out of sheer necessity. Which is by no means the same as suddenly realising #r1nt is here to stay bc it is claimed it does the job as well as on site interpreting. /3
Read 25 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(