Tech conferences: Why is nobody using an AI translator?

People love conferences for a reason. Navigating between in-person, remote, or hybrid setups, all of us can get a benefit we seek the most – traveling, keeping up with the latest trends, expanding our knowledge, or networking.

Conferences attract specialists from various backgrounds, but it seems that the tech industry is leading the conference game, with the IT sector alone holding 33% of the market share in the US. 

The tech industry is actually blooming all over the world. When technology experts arrive at global conferences, they get into the pool of diverse cultures and different languages. 

And for such distinct audiences, there is one central question: 

How can conference organizers ensure a first-class experience and efficient communication?

If you’re up to date with the new language trends, the first thought that probably pops into your mind is using machine translation. However, even if one of the main topics in the tech industry is artificial intelligence, technology conferences aren’t using AI translators, and today we will discuss why. 

The characteristics of a tech conference

There are hundreds of events for technology enthusiasts happening globally – just take Momentum, or IBM Think, for example. 

While conference organizers might be different, the event industry itself shares a lot of similarities.

Here’s a list of the shared qualities among modern tech conferences:

1. Virtual setup

Most of the tech conferences are currently hosted virtually on a remote conferencing platform of the organizer’s choice, which allows attendees to connect globally, forming a community of emerging experts. On the downside, online conferences provide less hands-on experience – a crucial part of tech conferences. But there’s always a bright side: people who have never participated in global conferences can now log in remotely. 

2. Complex topics

While some tech events are niche and specific, most of the conferences cover similar topics. World Economic Forum highlights some of the trends:

• Applied artificial intelligence, covering language processing, neural machine translation, and other topics;
• Process automation, including the Internet of Things (IoT) applications;
• Process virtualization, such as 3-D or 4-D printing;
• Future of connectivity, 5G being pre prominent example;
• Infrastructure, covering cloud and edge computing.
  

While the list could go on and on, we’d like to encourage you to think about these topics. Are you familiar with all of them? Probably not, and neither are we. That is the thing with tech conferences – they usually cover really complex topics.

3. High precision

Complexity is the thing that distinguishes tech conferences from other events. That’s why explaining difficult tech-based concepts in other languages requires high precision, a lot of attention, and focus. Something that AI technology is not capable of doing yet, and here’s why. 

Why is machine translation not a match for human interpretation?

In most multilingual tech conferences, event organizers stick to interpretation services carried out by professional conferencing interpreters. In this situation, real interpreters are manually transforming conference talks to the target languages of the audience. 

Below are the main differences between machine translation and human interpretation:

1. Accuracy

Machine translation may be a much faster way to deliver information to another language, but it is also less accurate than human translation. As we’ve mentioned before, accuracy is critical in tech conferences, and event organizers are not ready to risk the quality of their events going forward. 

It’s hard to predict whether machine translation will ever get close to the accuracy and consistency demonstrated by qualified interpreters. But there’s one thing we know for sure – the machine translation market will grow fast! Market analysts expect AI interpretation to grow by 7.1% in the next 5 years, which might bring a lot of technical improvements.

2. Context

Unlike a Math problem, language is grounded in context. Depending on the situation, speaker’s intonation, the dialect in a specific location, or the meaning of a phrase can be different. Those subtle differences are referred to as contextual errors. Contextual errors are also connected to accuracy loss. Most AI systems translate text without contextual information. It means that machine translation can’t predict what the speaker will say based on their verbal and non-verbal cues.

The only way to avoid contextual errors is to use human interpretation – our brain can interpret contextual information without any struggle.

3. Problem-solving

This one is simple: machine translation is currently not capable of efficient problem-solving. Interpreting at tech conferences doesn’t stop at delivering one speech from start to finish. It comes with many unexpected talks and panel discussions changing on the go. Unlike machines, humans learn problem-solving throughout their lives and can adapt to change very quickly.

4. Expertise and mistakes

Understanding the industry terms is vital in tech conferences. AI researchers claim that neural machine translation (NMT) technology is far from resolving its problems. NMT will make some of the mistakes that humans who have expertise in a specific field would never do, such as “misspellings and mistranslations of proper nouns or rare terms.”

Without any doubt, AI tools are the future of interpretation for general everyday situations we face when traveling to a foreign country or interacting with a stranger in the street. However, pure machine translation will never be the best fit for situations where precision and accuracy are rated above all. If anything, we expect machines to become great assistants for professional interpreters, helping them deliver the best experience possible.

‍