AI, translators, and job degradation

AI and the Changing Role of Translators

Translators are working in a changing and competitive field. New technologies, like AI, are causing their work to be dehumanized and their working conditions to decline. This may be the future for many other specialized professions.

Professional translators have been dealing with the impact of new technologies for decades. They have had to adapt to an uncertain and changing environment. Each new technological development provides useful tools but also adds complexity, dehumanization, and a loss of control over their work.

In translation, mastery of languages, cultural immersion, and knowledge of context are essential for accuracy. Translators rely on experience, sensitivity, and personal judgement. However, technology is being used to prioritize profitability over quality, sidelining the worker.

As AI capabilities expand, the world of translation is transforming. Other professions, including administrative personnel, auditors, lawyers, recruiters, managers, advertisers, analysts, journalists, artists, and creative professionals, are also facing similar challenges.

The Impact on Translators

There are approximately 640,000 professional translators worldwide, with three out of four being freelancers. This majority is experiencing a rapid, technology-driven decline in their profession.

Freelance translators can join translation pools of large companies after passing tests. These companies take a commission as intermediaries and can lower translators’ rates. Larger agencies, accounting for a fifth of the market, use technology to speed up work and lower costs.

Jean-Jacques, a translator with almost 30 years of experience, has seen his work conditions degrade despite embracing new technologies. He noticed the limitations of machine translation early on and watched agencies integrate neural network-based translation into their computer-assisted translation (CAT) tools starting in 2016.

How CAT Tools are Used

CAT tools segment texts into translation units, present the document in a grid format, and store translated segments in translation memories (TMs). The European Union institutions make their TMs publicly available.

CAT tools propose sentences from the translation memory to speed up work. Agencies use pre-translated texts to cut costs by paying less for sentences already in the translation memory. Rates vary based on the percentage of matches between segments.

This pre-processing has become standard practice, distorting the cognitive effort and creating a dehumanized work environment. Translators receive segmented files with machine translation suggestions, turning their role into revising and correcting machine translations.

Jean-Jacques explains that these tools don’t understand the text and may offer high-match translations that don't fit the context. He is paid less but has to correct mistakes and rewrite from scratch. Machine translation tools can also add or remove parts of a sentence, requiring intense focus to check every detail.

The Translator's Perspective

Rosa, a translator with two decades of experience, agrees that machines cannot replace humans and that machine translation is often inadequate. She enjoys working with direct clients but finds that agencies prioritize profit, treating translators poorly and demanding the most while paying the least.

Rosa describes translating segments in tiny boxes within a complicated system, which wastes time and energy. She is paid less than half her usual rate and accepts it out of necessity. The process involves countless administrative tasks, and errors can lead to penalties.

Automated platforms pay significantly less and distribute work through automated email alerts. Translators rush to translate available segments, and corrections can lead to disconnection from the system. Some platforms auction translations, with the lowest bidder getting the work, raising concerns about quality and translator satisfaction.

Neural network-based AI, which emerged in 2016, enables automated pre-translations and is used to replace human translators for predictable texts. José F. Morales notes that AI without human supervision may handle monotonous jobs but struggles with nuances and sentiment, impacting language quality.

Morales says that AI is causing strange uses of English to become normalized, creating a feedback loop that worsens over time. He suggests treating AI as a student that needs supervision to ensure good results.

Rosa emphasizes that machines can only replace humans for impersonal and repetitive texts. She worries about the future of writing if translators are squeezed to the point where their job becomes unviable.

Alina believes that AI is both a tool and a threat, learning from translators and potentially replacing them.

Seeking Solutions

Lindsay Weinberg and Robert Ovetz suggest rejecting AI interference and defending creative work. Translators may be seen as technophobic for resisting, but they are defending the quality of their working conditions. Translation requires cultural sensitivity and awareness of context, which resists automation.

Freelance translators are isolated, and professional associations lack the power of unions. Ovetz advises mapping out the work structure and identifying weak points in the supply chain to disrupt it. Translators can find allies among unionized employees of their clients and encourage them to change work assignments and payment methods.

Ovetz recommends accessing original texts to highlight flaws caused by automation and organizing around this to discredit the work process. Technology streamlines tasks, breaks them into components, and outsources or automates parts of the process, leaving the leftovers to human translators.

Jason Resnikoff argues that narratives of technological progress are a losing game for workers. Unions must reject employer-initiated changes without appearing opposed to progress and define their own concept of progress. He suggests keeping sight of a fair society where automation does not disrupt the labor fabric.

Elena, a translator with over three decades of experience, considers herself misemployed and appeals to trade unions to support “employer-less employees.” She argues that freelancers are wrongly categorized as companies and lack the rights and protections of salaried employees.

Elena believes that freelancers cannot make a profit like companies because their work trades time for money with physical limits. Employers are trying to remove human translators from the equation through AI, ignoring the need for human quality to ensure accurate and useful texts.

Ovetz insists on organizing and raising awareness about the weaknesses of this system, urging translators to inoculate newcomers against the challenges they will face and to get them onboard with organizing to avoid division.