Please enable JavaScript.
Coggle requires JavaScript to display documents.
Machine Translation - Coggle Diagram
Machine Translation
Why not?
-
-
-
"AI is bad"
AI has been in existence for as long as computer technology has existed. We have learned to live with many forms of AI that have come to be considered acceptable, such as smartphone voice assistants. As a society we have also adopted many forms of automation that have improved efficiency in their fields and become widely accepted, such as self-serve checkouts, automated elements of call-centre flows, and spell-checkers.
Generative AI and machine learning have been under scrutiny in recent times because of how they have been abused and misused (e.g. non-artists profiting from generative engines that were trained on other artists' work without their consent; machine learning being used to replicate the performances of actors without their consent; generative AI being used to produce works from scratch in their entirety without any human input). That is not what we want to implement in our team.
Translators who have moral objections to working with machine translation should have the freedom to choose not to do so.
-
-
-
-
Why?
MT, like CAT tools, has the power to automate the tedious, monotonous and time-consuming parts of job, leaving us with more time to focus on other areas of our work that we more productive, or that we enjoy more.
As part of a large, global business, finding ways to improve our efficiency and keep costs down helps the organisation as a whole.
Automated solutions help human workers to handle overwhelming workloads and deliver at the same quality as
MT is a proven effective measure to improve translation productivity without impacting quality of deliverables (when used properly, that is) — when RWS introduced machine translation to their workflows, productivity across their English Language Office increased by around 30%
MT is typically applied during the "Preparation" stage of preproduction, so it is a solution that fits readily into our pre-existing workflows with minimal disruption.
Why JCSE?
Our team is struggling under the current workload and this is likely to get worse rather than better.
If we cannot deliver on what is asked of us, other teams will likely employ other solutions such as asking members of their team to give up time to do the translation themselves, or using their own MT solutions. This takes control and agency away from us and increases the risk of low-quality translations circulating within and outside the company that might be unfairly attributed to us.
We are all in this together. Issues that affect our team affect the whole company. By improving our own processes, we help other teams to thrive as well.
Why now?
There is pressure on us as a team to improve our efficiency, and as other teams are finding ways to make AI work for them, other teams are likely wondering why we haven't done the same.
Many other teams in our organisation have found ways to use AI to expedite workflows (e.g. path-finding and bug detection in QA workflows) which has a knock-on effect on job satisfaction and the quality of our products, and we have an opportunity to learn from these other teams' successes.
Machine translation has improved significantly in recent years, and there are now many enterprise-level solutions readily available that can be tailor-made to fit the task at hand (DeepL, RWS Language Weaver, etc.)
-