I am a senior analyst for a global research firm, CSA Research. Although we don’t build the technology ourselves, we do examine the issues that keep enterprises from incorporating the full variety of human language into interactive agents and advise companies on how to bridge the gap between what they have and where they want to be. The challenge is in dealing with people in their own language and providing relevant information and data to them. We have focused on providing guidance for how to find or manufacture in-language training data or to bridge the gap using translated data. Most developers of chatbots and other intelligent agents tend to assume a monolingual world until they are forced to consider other languages. At that point they tend to see them as just a nuisance rather than an opportunity. However, our research has shown that about 75% of people won’t buy products outside of their native language(s), and that includes people who speak English really well (such as Germans or Dutch). They don’t want to have to use it, and this preference grows stronger the more intimate the brand experience is or in stressful situations. Gathering and leveraging multilingual data can help the billions of people who aren’t fluent in English or other major languages to participate fully in the online experience. If you assume that English is enough, you’re walking away from over half of the world and the business they can provide.
The telephone totally revolutionized the way we could communicate with people all over the world. But then came email and took it to the next level. And then came text messaging. And then came video calls. And so on…What’s next? What’s just around the corner?
In this interview series, called ‘The Future Of Communication Technology’ we are interviewing leaders of tech or telecom companies who are helping to develop emerging communication technologies and the next generation of how we communicate and connect with each other.
As a part of this series, I had the pleasure of interviewing Dr. Arle Lommel, a senior analyst with independent market research firm CSA Research. He is a recognized expert in quality processes and interoperability standards. Arle’s research focuses on technology, quality assessment, and interoperability. Born in Alaska, he holds a PhD from Indiana University. Prior to joining CSA Research he worked for the Localization Industry Standards Association (LISA), Globalization and Localization Association (GALA), and the German Research Center for Artificial Intelligence (DFKI) in its Berlin-based language technology lab. In addition to English he speaks fluent Hungarian and passable German, along with bits and pieces of other languages.
Thank you so much for joining us in this interview series! Before we dive in, our readers would love to “get to know you” a bit better. Can you tell us a bit about your ‘backstory’ and how you got started?
Icame out of an academic background (linguistics and ethnographic research) and I always envisioned myself becoming a professor and, if that didn’t work, I wanted to make musical instruments. As it happens, I’ve done neither of these things for various reasons, including the fact that instrument makers tend to be rather undercompensated. Instead, I did a stint at the German Research Center for Artificial Intelligence (DFKI) in Berlin, where I worked on the intersection of human and machine translation, then I joined CSA Research just before the current artificial intelligence (AI) hype cycle started in late 2015. There I’ve had the privilege to help some of the world’s leading companies succeed in their international efforts.
Can you share the most interesting story that happened to you since you began your career?
I was leading a portion of a European Commission-funded project and had just delivered some preliminary research at a conference that demonstrated pretty conclusively that a common method in my field didn’t work at all. After my presentation some researchers told me my research was uninteresting and not relevant to the field. They didn’t even engage with the substance of my argument, because to do so would have required them to face the possibility that they were wrong. However, later on a representative from a major automotive company came to me and told me that my presentation was the only useful thing he’d heard in two days of presentations. The researchers’ reactions just showed how self-interest can overrule anything else. Nobody likes to be shown that they have feet of clay, and it is easy to deny things, no matter how well argued, that work against the status quo. Without going into too much detail, I’ll just observe that major tech companies have since repeated the sorts of claims I had demonstrated are a problem and gotten major press coverage for them, which has created havoc for some in international communications as it has established truly unrealistic expectations.
Can you please give us your favorite “Life Lesson Quote”? Can you share how that was relevant to you in your life?
“Opportunity is missed by most people because it dresses in overalls and looks like work” (Thomas Edison). A lot of the work I’ve done with technology has come about because I’ve tried something that didn’t work and had to try again. Failure just means you’ve learned what doesn’t work and will be smarter the next time. I used to think that folks who got everything handed to them were lucky, but I’ve since realized they aren’t at all, because getting everything with no effort doesn’t begin to prepare you for any real challenge. Easy wins don’t prepare you for challenges at all.
None of us are able to achieve success without some help along the way. Is there a particular person who you are grateful towards who helped get you to where you are? Can you share a story about that?
There are really too many to single out just one person, but a single encounter stands out for me. Many years ago, I worked in a job that had asked me to compromise my integrity in little ways to get ahead. They weren’t big things, but the cumulative effect was that I lost my way a bit. I then went to work for DFKI in Germany and I was talking to my boss, Hans Uszkoreit, one day. Hans had spent time in an East German prison as a political dissident before he managed to escape to the West. He had been imprisoned for painting an anti-Soviet slogan on a wall, but what had really enraged the authorities was that he had used Western paint and no matter how many times they painted over it with Soviet paint, the slogan showed through. (I suspect there is a metaphor for life in that.) He then told me how they had tried to convince him it would go better for him if he just confessed where he got the paint, but they would have then held his concession over his head to turn him and get him to do what they wanted. When Hans told me this story, he paused and said that the evil genius of the Soviet system was that it asked you to compromise on the little things and then the big things followed along. It wasn’t until this moment — which I suspect Hans doesn’t recall at all — that I realized I had been compromising for far too long. Working with Hans ended up teaching me a lot and set me up for the position I am in now.
How have you used your success to bring goodness to the world?
If you speak English as your native language, it’s really easy to assume that technology just works, and that information is at your fingertips. You might use Google Translate on occasion, but if it doesn’t perform well, you’re probably not going to have much trouble. But now imagine you don’t speak English or one of the few languages that businesses typically target (French, Italian, German, Spanish, Japanese, and Chinese). You’re likely to be cut off from whole swaths of interaction, commerce, and information. My work has provided me with a platform to advocate for the value of language and international customers to many businesses. Of course, most of them go on thinking of language as a nuisance to be dealt with, but occasionally you see somebody catch the bug and really embrace it as an opportunity and investment.
Ok wonderful. Let’s now shift to the main focus of our interview. Can you tell us about the cutting edge communication tech that you are working on? How do you think that will help people?
I am a senior analyst for a global research firm, CSA Research. Although we don’t build the technology ourselves, we do examine the issues that keep enterprises from incorporating the full variety of human language into interactive agents and advise companies on how to bridge the gap between what they have and where they want to be. The challenge is in dealing with people in their own language and providing relevant information and data to them. We have focused on providing guidance for how to find or manufacture in-language training data or to bridge the gap using translated data. Most developers of chatbots and other intelligent agents tend to assume a monolingual world until they are forced to consider other languages. At that point they tend to see them as just a nuisance rather than an opportunity. However, our research has shown that about 75% of people won’t buy products outside of their native language(s), and that includes people who speak English really well (such as Germans or Dutch). They don’t want to have to use it, and this preference grows stronger the more intimate the brand experience is or in stressful situations. Gathering and leveraging multilingual data can help the billions of people who aren’t fluent in English or other major languages to participate fully in the online experience. If you assume that English is enough, you’re walking away from over half of the world and the business they can provide.
How do you think this might change the world?
It’s tough to think about changing the world, but even in the U.S. there is an urgent need for companies to engage with linguistic diversity. If U.S. Spanish were treated independently, it would be the eleventh-most-significant language in the world in terms of economic potential and another 12 languages that are as valuable as at least one of the European Union’s official languages. There are millions of people who would benefit from multilingual communications technology even if companies never go anywhere else. Think of how ubiquitous interactive digital agents have become. Just the other day I had to interact with a large retailer’s support department, and I had to go through a chatbot to actually talk to a person. But if I didn’t speak English, I wouldn’t have been able to even start that conversation. This leaves many people at a profound disadvantage that technology can help solve.
The “universal translator” or Babel fish is a trope of science fiction that allows TV and film directors to avoid having to deal with language (Arrival being the rare exception), but the thought of being able to talk to whoever you want to, whenever you want to, without worries because they happen to speak Czech, Chuvash, or Chinese — but not English — is tremendously appealing. Some of that can be automated, but a lot cannot. However, by increasing the expectation of translation and in-language experience, our research, and the work of our clients, can really improve the world. Incorporating this language experience more widely into conversational AI will create a lot of goodwill and help customers — particularly those in stressful situations — interact with brands more effectively.
Keeping “Black Mirror” in mind can you see any potential drawbacks about this technology that people should think more deeply about?
The recent rise of machine translation (MT) as perhaps the flagship technology of recent developments in AI offers a lot of benefit, but when people accept it uncritically — encouraged by credulous press accounts of it reaching the same level as human translators, or even beating them at their own game — it can lead to real problems. ProPublica reported that US border agents have relied on MT to interrogate detainees or examine their social media histories. It is vital to remember that AI isn’t actually intelligent in the sense humans are: It doesn’t understand what it ingests, and certainly cannot deal with humor, sarcasm, or even casual speech. It needs a warning label that states that you use it at your own risk and that you cannot rely on it when dealing with real human needs. Even if businesses aren’t going to deport people based on misunderstanding, there is a real risk that chatbots, unless backed up by human attention, could do things that would affect brands or customers in negative ways. For example, imagine a chatbot with imperfect training data that tells your customers something that applies to competing brands or that repeats something offensive in its training data that never showed up in testing.
Was there a “tipping point” that led you to this breakthrough? Can you tell us that story?
One of CSA Research’s clients was a large multinational tech company, a real leader in the development of conversational AI, and the AI team members discovered quite quickly that while they could develop 10 or 20 intelligent services in a year, they were struggling to get a single non-English one out every six months because the data just wasn’t there. This particular company does business in most countries in the world and supports around 70 languages in its consumer products, not to mention its business and government lines. In other words, is literally has thousands of years of work using present approaches just to catch up to where they already were in English with their intelligent agents, not to mention full-blown conversational AIs, and that’s all without even considering everything else churning out at an ever-increasing pace. It was this realization, that even the market leaders in this area were facing seemingly insurmountable challenges, that led us to develop an approach based on the Three C’s — collection, curation, creation — that helps enterprises understand what data they have, what they can translate, and what they need to create from scratch to get there. It still isn’t easy, but we find that most companies have been discarding relevant data for years because they aren’t aware that they are generating it. Getting them to understand the value of customer interaction data and collect it can make a huge difference going forward.
What do you need to lead this technology to widespread adoption?
Two primary obstacles stand in the way. The first is lack of awareness. Very few organizations have ever thought about the strategic value of language and the need to collect data from outside their home market. They also tend to gather data only in their home market but let others languish. As a result, their CMO may come up with a great approach for the headquarters country and then invest nothing to make it real outside that market. The second is that the developers of chatbot frameworks — and semantic technologies more generally — often operate as though English were the only language in the world. We need to see standards, such as Schema.org, and developers really need to incorporate language-oriented features into their work from the ground up. Bolting them on after the fact always leads to sub-par results. Similarly, companies need to plan for market and language needs from the beginning rather than trying to retrofit their products after they are mostly built. That will require companies to change how they invest in technology development and standardization and start to remember the rest of the world in everything they do.
The pandemic has changed so many things about the way we behave. One of them of course, is how we work and how we communicate in our work. How do you think your innovation might be able to address the new needs that have arisen as a result of the pandemic?
Perhaps the most striking aspect of the pandemic from our perspective is that it has decoupled residence, work, and commerce as geographical factors in surprising ways. One of CSA Research’s clients told me about how one of his company’s own customers — a large UK supermarket chain –suddenly found a pressing need for online support for Urdu, the variety of Hindustani spoken in Pakistan and one of the hardest languages to display on-screen. It turns out that lockdowns had made it really difficult for customers with limited English who previously could walk into the store and pick up a cabbage or some other items because now they had to read an online order form and know the names of grocery items in English. They found this hidden need for an “exotic” language in the heart of England, and it was far from trivial to support this. These needs are only growing in importance.
Companies have also found that if they don’t have big offices in the city center, they might as well look for the best talent they can find, and that might be in another country. If they cannot support workers elsewhere in the world, they are limiting themselves unnecessarily. Sales, marketing, and support efforts increasingly rely on automated agents, which have become more crucial as they have replaced face-to-face interaction. But when you are using your phone or tablet for communication with your work or with a brand — or with a utility or a government agency — you want to do it in your own language. Building agents designed to be multilingual from the start opens you up to your customers and extends your reach to the world in ways that make your business more resilient and responsive in the face of challenges. Our clients are also struggling to add language support for live, online events, but discovering that they are more effective in their sales and marketing efforts now than they were when they relied on large, in-person trade events. They now want to extend this to all areas of marketing but are just beginning to realize how challenging this will be.
Fantastic. Here is the main question of our interview. What are your “5 Things I Wish Someone Told Me Before I Started” and why? (Please share a story or example for each.)
- Be your own toughest skeptic. I don’t mean that you should treat yourself negatively, but if you subject your own ideas to tougher criticism than anyone else will, you will be prepared for any challenge that comes. Any time I go into a presentation or a pitch, I try to think what someone who disagrees with me will say and about how I will answer that. It helps improve my ideas and also means that I’m seldom caught off guard. For instance, I recently did a presentation to one of the European institutions and at the end one highly skeptical individual basically said he thought my presentation was completely off base and then went on to explain why in terms that could have been devastating if I were not prepared. Because I’d already thought about his objections and could contextualize them without feeling threatened, I was able to come back with a strong response that garnered support from others.
- Be grateful for your honest critics. We live in a time when it seems a lot of people see disagreement as bad thing and any criticism as a sign that you are their enemy. But at the end of the day, listening to — and understanding — your critics is the only way to avoid walking naïvely into foreseeable problems. More than once, someone has shown me something I desperately needed to know that I would have missed if I had just shut off disagreement. And more than once I’ve found that we were actually in agreement at a deeper level and that we could find a compromise because we didn’t see each other as the enemy. This doesn’t mean you shouldn’t passionately defend your positions, but always be open to what you can learn from others.
- You’re probably wrong about most things you think, and that’s OK. Think about it for a minute: How likely is it for any given topic that you (a) know all the possible solutions to a problem and (b) have picked out the right one? Just as a matter of the odds, on any substantive question, you are therefore more likely to be wrong than right. Now multiply this by hundreds or thousands of topics and it’s clear you’re going to be wrong most of the time and that you shouldn’t trust everything you think. But this isn’t a negative: If you’re likely to be wrong, then you have the freedom to admit that you were wrong when something better comes along. People who won’t admit they are wrong and stick to their conclusions when they should drop them go on to create a huge blast zone of damage.
- Pay attention to the small things. If you get the details right, the whole will come along, but if you ignore the details to focus on the big picture, you’re likely to find that neither the whole nor the details are right. I remember one particular analysis project I was working on where I wasted weeks of effort because of a seemingly small problem that I kept meaning to get back to, only to find that it actually hid a major one that required me to start over from scratch. This applies to not only work but life in general. If you sweat the small stuff, you’re usually getting the big stuff at the same time. The challenge is to know which details to focus on, so you don’t just get lost and flounder. If you’re a “big picture” sort of person, work with people who focus on the details and who can keep you grounded.
- Know when to walk away. Most tasks reach a point of diminishing return. Although perseverance is admirable, the sunken cost fallacy leads to bad outcomes. Consider Theranos, which was started under the seemingly revolutionary premise that you could run all sorts of medical tests with just a few drops of blood, even though a beginning analytic chemist could have told them this was fundamentally impossible. Even when it should have been obvious it would fall short; people wouldn’t admit that and at some point, it went from dogged dedication to fraud. Even if most of us will never waste billions of dollars in this fashion, being able to let something go removes a lot of stress and frees you up to move onto more productive things.
You are a person of great influence. If you could inspire a movement that would bring the most amount of good to the most amount of people, what would that be? You never know what your idea can trigger. 🙂
If I could inspire a movement, it would be to make it easier for anyone with the desire and interest to go live a year or two in another country. My time living outside the US — in Hungary and Germany — was tremendously influential in my life and it matters every day because it changed who I am. My children spent much of their formative years in Germany, and I’d like to think it will help inoculate them against some of the darker impulses we see in politics and society today, ones that couldn’t exist if people had direct experience with people who speak, act, and live differently than they do, but who can still be friends.
How can our readers further follow your work online?
Most of my writing is for our clients, but my posts in CSA Research’s blog are a good place to see what I’m doing.
Thank you so much for the time you spent doing this interview. This was very inspirational, and we wish you continued success.
About The Interviewer: David Liu is the founder and CEO of Deltapath, an award-winning unified communications company that liberates organizations from the barriers of effective communication. Liu is known for his visionary leadership, organic growth strategies, and future-forward technology. Liu is highly committed to achieving a greater purpose with technology. Liu’s business insights are regularly featured in Forbes, Entrepreneur Magazine, Tech Crunch, and more.