This blog post was originally published on Medium.

Since the first Large Language Models (LLM), like the GPT became available I started wondering about the implications they will have for the content creation and translation business. Here are a couple of things to ponder as we embark on this new journey.

How did Machine Translation disrupt the translation industry in the past?

Looking back in time, we can take some cues from the way Machine Translation (MT) has disrupted the human translation business. Despite the worries of many, machine translation has democratized access to translation and made it much more affordable. This resulted in the automation of translation of low-profile content, for which previously human translation was not an economically feasible alternative. This in turn boosted the amount of content that was translated, ever more increasing demand for translation services

Looking at the analysis by CSA Research from a few years ago, the language industry market size and Compounded Annual Growth Rates (CAGR) have been growing steadily. In late 2016 Google announced an amazing technological breakthrough — the introduction of Neural Machine Translation that delivered exponential gains in translation quality.

Did it impact the industry? Sure, but looking at the growth data, not really. Any negative impact that neural machine translation might have had on the industry was offset by the increase in the amount of content being translated. For translators, machine translation became an efficient tool, just like Computer Aided Translation (CAT) tools and Translation Memories (TM). It helped translators automate the mundane translation of low-profile and allowed them to focus on higher quality and higher visibility content where they could add more value.

How does ChatGPT change content economics?

While machine translation technology disrupted the translation business by increasing the efficiency of translation work and in some cases eliminating the need for human intervention, disruption coming from Large Language Models will have different implications.

Large Language Models will likely lead to another technology paradigm shift in machine translation and generate another leap in machine translation quality output. LLMs will get integrated into MT engines or even replace them. This will further boost the forces at play, enabling the use of machine translation in more use cases and content types.

Large Language Models and generative AI can create content that resonates — though not always free of errors, it successfully delivers the message. And that might be enough for most content. Moreover, it can do so in dozens — if not hundreds of languages — and at a fraction of the cost of regular copywriters.

Just like in the case of machine translation technology for the translation business a few years ago, generative AI will create a paradigm shift for anyone who is in the content creation business — to a certain extent everyone.

This will be of significance for content with short life spans that needs to be created and delivered to audiences quickly and will lose importance within hours or a day at best. With such a dramatically reduced cost of creating original content directly in the target language, generative AI creates a viable alternative for translation.

Alternatives for content managers

The benefit of creating versus localizing is that content created from the scratch in different languages is that original content will rank better. As the generative AI models have been trained on the content available on the Internet, from the scratch it will use more appropriate terms that are used by real people.

With the advent of ChatGPT and similar generative AI technology content, managers are getting more options to choose from when building their content strategies and deciding on the best way to get content in front of users.

Here is an overview of different alternatives that content marketers might consider when deciding on their content strategy. With ChatGPT, at current price points and the quality that will only improve in the future the question stands: does it still makes sense to translate content?

There is no easy answer. It depends on the content and its purpose and how deliberate your content strategy is. In a well-thought-out content strategy, ChatGPT can play a significant role and help automate a portion of your content creation efforts enabling content managers to get more content in front of their users.

ChatGPT will also help marketers and content managers personalize their content and tailor it to ever smaller audiences.

So the question stands — why translate content when you can generate it?

While many will be tempted to embrace ChatGPT head-on, I advise caution. Automatically generated content looks good and will work in many use cases but it comes with bias or maybe just inaccurate. Read more on the risks of using ChatGPT in my other blog post.

Generative AI, will not replace humans but will serve, as a productivity tool that allows content creators to quickly generate good drafts and allow them to shift focus on adding more value to the content, rather the writing it from the scratch. It will enable increased content personalization across languages — to a level never seen before. Over coming years we will experience deluge of automated generated content but this time it will be smarter, more tailored to our needs, smarter.

Humans will continue to play important role in the content value chain and their focus will shift to higher value content where they will continue adding value and improving customer experiences.

Notes on the above calculations:

1. This is my own research and by no means should be treated as a recommendation.
2. Prices have been calculated on 500 words long piece of generic business content translated
3. The price ranges are indicative of average mid-price tier languages like Spanish or French.

Sources for cost estimations:

1. Machine Translation — https://translate.google.com/
2. Professional Translation — https://www.getblend.com/
3. Transcreation — https://www.translationpartner.com/cost-of-translation-services/
4. Copywriting — https://prozely.com/copywriting-rates/
5. ChatGPT — https://www.ciocoverage.com/openais-chatgpt-reportedly-costs-100000-a-day-to-run/
6. ChatGPT + Proofreading — https://www.magnumproofreading.com/post/how-much-should-i-pay-for-a-proofreader