We previously discussed the future of Artificial Intelligence Translation (also known as Machine Translation) and the potential impact on how we deliver language services in the translation industry. One of the article's main points was using artificial intelligence (AI) responsibly. When AI is used responsibly with the proper quality assurance steps by human translators and editors, it can be a great asset in improving your business' efficiency.
Here are four use cases where using AI translation would benefit your business:
1. Handling customer service requests
Customer service departments are flooded with emails, calls, and chat requests. Unfortunately, many of the recommendations tend to be spam or illegitimate requests. Timely processing of these requests is essential since important customer service needs are included in these messages.
Setting up a workflow to handle the automated translation of these messages decreases the time required for a response and helps eliminate illegitimate requests. AI is set up to handle the influx of messages and queue up the results for expedited human review. Valid requests can be adequately routed, while spam can be deleted.
This workflow allows for the inexpensive and speedy processing of requests and helps companies focus their resources on legitimate customer needs.
2. Internal procedures and processes (low-risk procedures only)
Many companies host internal knowledge bases with hundreds of articles. If the subject matter is relatively low risk, a machine translation plus post-editing (human review) workflow is appropriate. This type of workflow also allows for deeper automation by connecting a workflow to the customer's content management system.
In a recent project, we created a connector from our translation management system to our customer's SharePoint instance. We now look to the database daily for any new or modified content. The connector processes the articles and submits the content for translation. Machine translation handles the first translation pass, while a human editor does the final quality assurance.
This process works because we have an extensive translation memory database for the customer made up of all of their past translations. This process allows the workflow to leverage previously human-translated content first, and then the machine translation engine provides any missing content. The human editor adjusts any incorrect translations, and the customer has an acceptable translation for use internally.
3. Research projects
Customers sometimes face large volumes of translated materials without completely understanding the significance or relevance of those documents to their business. This sometimes happens during the acquisition of a company outside of the United States, due diligence in the acquisition of a company outside of the United States, competitive research, or even the receipt of an extensive RFP written in a different language.
These projects typically start with a small set of documents, and additional documents are added over time. Connecting an automated workflow like a monitored FTP folder or file-sharing folders like Box or ShareFile allows easy requests for translation and scheduled pickup and delivery of content. The translation process automatically kicks off when a file is added to the folder. Some advantages of a workflow like this include monthly billing, allowing for different workflows (MT only, MT plus human review, etc.), and adding multiple file types and language combinations.
4. Large volumes of technical documentation
Customers who handle large volumes of technical documentation for low-risk product lines could benefit from an AI-boosted workflow. This is especially true if there are large amounts of repetitive text across their documents. An MT plus PE (post-editing) workflow could significantly reduce time and cost while maintaining a decent quality level. Use this process cautiously on high-risk documentation like user manuals for medical devices or heavy equipment.
Final thoughts
The bottom line is that when to use AI to boost a translation or language services workflow will rest on the customer's comfort level with potential issues or errors. A good AI workflow will have a feedback loop to capture errors and train the engine to prevent future mistakes. However, the potentiality for errors exists and should be well understood by all parties. Your language service provider should be completely transparent about the process of creating your translations. Your first step should be to thoroughly discuss how AI can help your situation and how it could potentially cause issues.
The language services industry will continue to grow. The thirst for cheaper and faster translation will increase, increasing demand for AI-boosted workflows. Without a doubt, the market for effective human translation or input from human translators will grow along with these conditions. The challenge for our industry will be to balance the resources and continue to use computing power responsibly to benefit the end customer and the vast network of human linguists.
Comments