My ChatGPT use cases

To get our weekly posts directly in your inbox, sign up for the Training for Translators mailing list!

Before we get going with this topic, I realize that people have a huge range of feelings about AI in general, and ChatGPT specifically. If you hate AI and don’t want to read a word about it, just skip this issue of the newsletter! I tend to feel that:

• AI is a tool that I want to know about, and that I’m interested in using when it’s helpful
• I hate certain aspects of AI: the environmental impact, the “better to ask forgiveness than permission” model of copyright infringement, etc., although I realize that these are also nuanced issues (a presentation that I’ll reference below, mentioned that if the massive electricity needs of AI end up driving demand for larger-scale renewable energy facilities, that could actually be a good thing)
• I also find AI useless for writing or translating from scratch. I find its writing style robotic, and it makes mistakes. I have the paid version of DeepL and I use it quite a bit for interpreting prep, but I never use AI to create a translation or something I’m writing from scratch
• My reaction to, “But AI is so great for automating boring, repetitive work,” tends to be, “But nothing about my job is boring or repetitive…that’s what I love about it!”

Enter the techforword AI in Translation Summit

The beta version of ChatGPT came out almost exactly two years ago (November 2022). I honestly hadn’t used it very much, until a couple of months ago, when I started hearing rumblings from people I trust, with specific ChatGPT use cases that made sense to me. A couple of other not-super-techie translators told me they found it helpful for complicated research questions (always check the sources!). My daughter (an engineering student looking for a summer job!) told me that she uses it to write LinkedIn connection request notes, because it can comply with specific character counts, and tone instructions (make this more formal, make this less formal).

The real tipping point came when I taught a class for someone else’s online training platform, and the organizer came back with what I thought was a fantastic title for the session. When I complimented them on it, they quickly responded, “Thank ChatGPT!” It was definitely time for me to try it for myself.

And, as luck would have it, Nora Diaz and Josh Goldsmith’s techforword AI in Translation Summit was coming up. Disclosure here: I don’t do affiliate marketing deals, and I pay for my own techforward insiders membership (well worth it, IMO), but they did give me a free Power Pack registration to the Summit as part of the compensation for a class I taught for them.

In watching the Summit presentations (particular shoutout to Bex Elder’s “How to use AI to help you create great content faster,” Veronica Hylak’s presentation on the environmental impacts of AI, from whence comes the renewable energy reference above, and Riccardo Schiaffino’s presentation on Perplexity as a research tool), I got some ideas of how ChatGPT could work well for me. Which brings me to…

How I’ve been using ChatGPT lately

1. Interpreting prep: Asking ChatGPT to predict what a speaker is going to talk about, based on the presentation title. I have the free version, so I don’t put anything remotely confidential into it. But I was recently assigned to interpret a speech on the topic of “The erosion of trust in public institutions in X country,” and the slides were pretty bare-bones. I asked Chat GPT, “I’m an interpreter, and I have to interpret a speech on the erosion of trust in public institutions in X country, can you think of some recent news events that the speaker might mention?” It came back with three suggestions, and the speaker mentioned two of them.
2. Jazzing up presentation titles. I’m not the most creative writer. What can I say…I’m an Upholder, I’m good at getting things done and ticking boxes, more than thinking outside the box. I asked ChatGPT to suggest titles for a few presentations I’ll be giving in the next few months (i.e. “I’m giving a presentation for beginning translators and interpreters. I usually call it “Getting started as a freelance translator or interpreter,” can you come up with a more interesting title?”).  I ended up using ChatGPT’s suggestions for all four of the presentations I asked it about.
3. Complex research questions. This “search engine on steroids” feature isn’t anything you can’t do yourself, but Riccardo’s presentation gave some examples where an AI tool can definitely save you time, just make sure to check its sources (all AI tools should provide the sources if you ask them to). Riccardo’s example was something along the lines of, “How are the child’s preferences taken into account in child custody cases in Italy versus in the U.S.” This led me to ask it several complicated research questions of my own: What are the parameters on non-citizens being self-employed in Switzerland, and do the parameters differ if that person is or is not a citizen of an E.U country? (This was a literal “asking for a friend,” I’m a Swiss citizen!). Note: Riccardo’s presentation and notes are available here, on his website.
4. Rewrites for tone or length. I stole my daughter’s idea for a LinkedIn connection request note (make it 200 characters and less formal), and I had it condense a couple of things that were too wordy. It also gave me some good wording suggestions for a translation I’m working on (“What’s another way of saying ‘seen the light of day’?”) in which case I used its top suggestion, “come to fruition.”

The AI landscape is relatively chaotic at the moment, and we really don’t know what’s going to turn out to be the new normal, and what’s going to be hype. As I’ve mentioned before, I really enjoy the daily newsletter The Neuron. I know…who wants another daily newsletter, but it takes literally five minutes, and I read it while I’m drinking my coffee and coming to my senses. Amid all of this uncertainty, hopefully my ChatGPT use cases have given you some ideas!