Article by Thomas Tolnai
This year, we wrapped up TranslateCluj with a simple challenge. Following all the information shared and the skills acquired during the two conference days, we asked our attendees to think of one thing they’re going to
- Start doing
- Stop doing
- Continue to do.
The best application of generative AI in our opinion is ‘assistant work’. If you’d like to learn more about how to turn current AI tools into reliable assistants, follow us to find out more about our upcoming online event, Straight to the Point, on May, 17th.
This article is about the one thing we, the organisers, started doing. And that thing is integrating AI in the organization workflow of the conference. At this point, this sounds like a cliché, but bear with me for a minute. Those of you who have been attending our conferences and following us on social media have probably realised by now that we are not early adopters. Not that there’s anything wrong with the concept, it’s just not the way we are. So when generative AI became a thing a couple of years ago, instead of immediately jumping on the new hype train and being the first to experiment, we took our time, a bit defensive, a bit skeptical and waited for the true innovators to share their results. About three years later, we organised an entire conference around generative AI and understood how it could fit into our daily work routine, where it can be very useful and where it is still lagging behind.
Contrary to what many of the translation industry stakeholders are preaching these days, we don’t think AI is good at translating, at least not as far as client-facing, high visibility, high risk, meaningful content goes. Sure, MT and LLM translations have their use cases, but I won’t insist on them. There are plenty of articles about that online. In terms of (copy)writing, though, I will say that not all copy is born equal and that not all texts require and deserve our full ‘human’ attention. In fact, for some of the English content we usually create (keep in mind that none of us is an English native speaker), AI apps have proven useful and have saved us a lot of time. For plain, factual announcements, for summaries and event take-aways, the custom AI assistants we created in Claude and ChatGPT have turned out to be pretty good. We’re aware that the output is fairly generic, but… these things are maybe supposed to be generic? There isn’t much you can innovate about announcing a registration deadline, a new speaker or topic. The stress falls upon information, and for that genAI is great.
What AI tools have really helped us with in the past month is business intelligence. The possibilities here are endless. It is, in fact, within this use case that we had our a-ha moment.
Unsurprisingly, then, we started using AI for writing. But this is not where the true value of these new tools lies. What they have really helped us with in the past month is business intelligence. The possibilities here are endless. It is, in fact, within this use case that we had our a-ha moment. You see, TranslateCluj is still a side project, you can call it a hobby, if you will. And as such, while we do hire some, and see the real value of, human consultants, there is simply no way we can afford the services of professionals for every chapter of our business. Instead, we started relying on the subscription-based ChatGPT and Claude apps and thanks to them, we’ve been able to generate meaningful roadmaps, communication plans and even blueprints of business development strategies. Just to give you an example, I was looking for a tool to assess a business idea. In the first step, I asked ChatGPT to find me several appropriate tools — this is how I discovered the concept of lean canvas. The next step was to create a custom GPT for TranslateCluj and apply the lean canvas model on our ‘business idea’. In a matter of minutes, I had a blueprint, a place to start. No, we don’t rely blindly on the results generated by the AI and we don’t follow all suggestions. But they’re a great place to start and they can inform our future decisions.
To sum up, the best application of generative AI in our opinion is ‘assistant work’. These apps can become meaningful assistants, highly useful either for repetitive, time-consuming tasks (think terminology extraction, source text analyses, corpus and file management for translators and interpreters), or for research, brainstorming and business planning. If you’d like to learn more about these tools and would also like to turn them into reliable assistants, you should definitely sign up for our upcoming online event, Straight to the Point, on May, 17th.
No AI has been used for writing this article. Everything in here is 100% human 😛.
