With increased automation, software reaches decision-making capabilities previously thought to be confined to the human brain. No profession is exempt, except when rendering these services requires a capacity for intimacy, compassion, and emotional response – for now.
Translation, for example, is one of the domains where developers have tried since 1951 to render text from one language to another, so far with limited acceptable results.
67 years later, we are at the point where the translation of technical documents can be automated fairly well. On the other hand, we still get oddly surreal “translations.” Some users, for example, once typed random characters into Google Translate and the Thai translation read: “There are six sparks in the sky, each with six spheres. The sphere of the sphere is the sphere of the sphere.”
One of Google Translate’s most spectacular fails went viral after a user filmed his attempt to translate two Japanese characters.
The video, viewed nearly 4 million times, shows the user entering the two characters repeatedly and noting the increasingly bizarre translations Google comes up with each step of the way.
Things go completely off the rails when he gets back translations like “Transportation Eastern maple Egg bag,” ” of deep-sea squeeze trees,” and for good measure, ” EGUEEGEGUGE deep-sea EEGEGEGYE EGGTAG.”
We will find such examples for a long time. My prediction is that machine translation will not pass the Turing test in my lifetime. Notwithstanding the fact that some human translations are not distinguishable from that of a machine, a machine will not have the ability to translate high-touch content equivalent to, or indistinguishable from, a human translation before I die.
But does it have to in order to kill your job? The answer is: yes.
Joseph Weizenbaum made this case in 1966 with his program ELIZA, an early natural language processing computer program that gave users an illusion of an actual conversation. It was one of the first chatbots. Its most famous script, DOCTOR, was so effective that individuals initially thought it had human-like feelings. In fact, it only used rules within the script to respond with non-directional questions to user inputs, such as:
Many users believed that the program could aid doctors in working with patients. Some even feared psychologists could go out of business.
ELIZA’s weakness was that it was unable to contextualize anything. And while latest innovations enable applications to respond to a person’s emotional state by combining voice with face recognition and emotion sensing, it’s not enough. Responding to a client’s need for relevance, belonging, personal growth and giving back will continue to be the advantage of warm-bodied project and customer service managers.
Coaches and psychologists are not just looking at words. They need to interpret a person’s body language, how a person relates to past events and trauma and meets basic needs. Great life coaches identify limiting beliefs and see unproductive patterns in their clients’ behaviors. They understand their clients’ world and emotions first, show different ways of dealing with life challenges, connect behavioral change to something higher than their client’s self to make it last, and show them one small change that helps them make a breakthrough each time they meet.
The death of the file pusher
As translators and localizers, we can learn from that. For one, machine translation will be of little use for content that requires emotional and cultural context. At least for quite a while. Second, while many tools can fully automate most transactional tasks of a localization project, they cannot respond to a client’s need for relevance, belonging, personal growth and giving back.
Yes, file management, email writing, reporting, quoting, scheduling, and many other tasks can now be fully automated. And if you are more of a file pusher than a neuro-science psychologist to your client, your job will be axed sooner or later. The picture below illustrates an actual implementation of a localization workflow in which a project manager is only needed to handle exceptions or when s/he needs to make a judgment call.
Localization professionals that can apply soft skills will always be in critical thinking, interpersonal interaction, consumer preference analysis, communication, team building, leadership, and social and cultural awareness are most needed in the future.
It’s about people and relationships
Being a pioneer in automating project management (I have done this for 20 years now), I have never found that technical shortcomings of translation automation are the problem. People are.
I spend most of my time coaching managers that automation is not just about processes and hand-offs, but also about people and relationships. Automation in localization fails because the pain of change is stronger than the pain of working the way things currently are. Coupled with the real possibility of automation killing jobs, there is often little incentive for teams to change.
There are two basic forces that determine our behavior: 1. the need to avoid pain, and 2. the desire to gain pleasure. The need to avoid pain is always the stronger one. That’s why well-meaning managers experience push-back from their stakeholders when they introduce new ideas with the best of intentions of building a better future. The big two questions in everybody else’s head is: ‘Will I be good enough, and if I fail, will I still be loved or respected?’ Their own future, in their mind, may not be so bright after all.
Empower – not limit – yourself
Successful localizers break unhealthy responses to inevitable automation by turning their limiting believes into empowering ones. Instead of telling themselves that process automation will take away their jobs, they create empowering statements, such as: ‘Automation allows me to eliminate all the clutter and repetitive work – and I have all the knowledge, experience, relationship skills and talents to now deliver more value to my clients than ever before.’
And then they train themselves in using their natural capacity for problem solving, adaptation, and resilience and cross-disciplinary solution building.
How much time do we have? It took six decades for laborers to resettle and start to win higher wages once factory automation took hold in the 19th century’s Industrial Age. How long localizers have in the age of automation is not clear, but the transition has started. Estimates are that in the next 15 years, 38% of American jobs would be lost to robotics and artificial intelligence; Germany would lose 35%, Japan 21%, and the UK 30%.
According to a report from consulting firm PwC, administrative and support services are more at risk than human health services, social work, and education. My case in point: If you spend much time pushing files or run a translation sweatshop you are already automated away. You need to act now.
To speak with Stephen Hawking: 1N73LL1G3NC3 15 7H3 4B1L17Y 70 4D4P7