Survival:AI, the Desktop Publishing of the 21st CenturyAdopting a new technology is not necessarily a sufficient strategy

William Lise (October 25, 2023)

For Japanese-to-English (or other) translators who think the key to surviving the translation agency MTPE business model is to adopt machine translation, here is an example I vividly recall of how adoption of technology by translators is not necessarily the silver bullet. The similarities to the current situation with the agency MTPE business model should be apparent.

It was the late 1980s in Japan and the Bubble Economy was blasting full-steam ahead. Bubbly people were bubbling away, buying condos they didn't need with money thrown at them by banks, and people thought it would never end. The folly of that expectation was to be demonstrated in just a few years.

Laser printers were extremely expensive and reasonably priced inkjet printers had not yet appeared. Some freelance translators proudly proclaimed on their profiles in places like the membership directory of Japan Association of Translators that they had a laser printer and others, probably more honest but a bit mysteriously, were proudly revealing that they had “access to a laser printer." I guess their employers were not reading the membership directory.

WYSIWYG displays had appeared, and significant numbers of translators had started selling DTP (desktop publishing) services with laser-printer output. How effective was that in the long term?

Potential clients also soon bought laser printers if they didn’t already have them, and here in Japan the replacement of domestically developed personal computers with proper Windows or Apple computers combined with laser printers quickly to obviate the ordering of DTP tasks from translators. Not too soon afterward, just providing a client with data became sufficient. DTP became much less special and valuable as a service.

The effectiveness of adopting a technology for one step in a multi-step production process is greatly reduced if your potential clients adopt the same technology and can perform that process step separately themselves, without spending nearly as much as vendors thought it was worth as an outsourced service for that process step.

And here we are almost four decades later, and translation agencies have adopted AI translation.

Some translators are thinking that they should simply adopt AI themselves, but they appear not to realize that most of them don’t have customers that will order from them the entire translation process, including the AI translation step.

The agencies certainly won’t be those customers, and agencies are already moving away from humans except for the post-editing process step.

For translators, AI machine translation promises to be the DTP of the 21st century—a dramatic advancement that their clients are adopting as a separable process step that need not be outsourced. But it is a bit worse this time. because the technology is not providing an ancillary service such as DTP, but rather the core service in the translation process, translation itself. And the clients are ahead of the translators in adopting a technology this time, but it matters little. Once they adopt it, what’s left for the translator is, well, just the other steps of the process, and they are neither enjoyable nor lucrative.