The new ChatGPT conversational model, developed by OpenAI, has quickly become a web necessity. The prototype’s capabilities are simply immense. You can ask ChatGPT almost anything and the programme will answer you almost immediately.
Working in the same way as Google Translate and Deepl, ChatGPT is also able to translate texts for you. But what about the overall resulting quality? Can ChatGPT translations be detected by Google and lower your SEO rating? Is confidentiality taken into account?
Let Milega e-translation agency tell you more.
Deepl or Google Translate equivalent translations
Yes, it’s true, ChatGPT can translate your website. But, despite the attractiveness of such a solution, it’s not a particularly good idea. ChatGPT machine translation is basic to say the least. It’s equivalent to other automated translation engines.
If you ever decide to have your website translated by ChatGPT, your content will certainly be translated for free, but it will also be full of grammatical or language errors. It will not be perfectly adapted to the target language of the chosen country.
Furthermore, ChatGPT is generally not able to correctly translate local expressions or language subtletiesn in others words, content localization. Consequently, your website texts will be of extremely poor quality. Your visitors will not have a pleasurable browsing experience, and parts of your site may even be incomprehensible.
Google identification of content generated by ChatGPT
In the same way as automatic translations, Google has made it clear that its algorithms are able to identify content that is 100% GPT generated.
In terms of SEO, Google’s algorithms penalise automatically generated content. GPT content will immediately lower your site in the search results. It is therefore particularly damaging for your digital strategy to implement automatically translated content into your site.
Also, while there is no clear evidence that Google cares about language models like ChatGPT, there is some form of competition between these two companies, especially in the AI domain. It is therefore likely that the Google search engine will harshly penalise content translated and therefore generated by ChatGPT.
ChatGPT Translation: Privacy and Data Protection
State-of-the-art Neural Machine Translation (NMT) models are better than general Large Language Model (LLM) models like ChatGPT. And this is for three very good reasons: quality, data privacy, and deployment options.
Quality
NMT models were designed as predictive models. They are optimised to be as accurate as possible. In addition, in order to generate additional quality in a given field, adaptation (machine learning) means they are constantly evolving. Conversely, how a shared generic AI model like GPT can be adapted has not yet been clearly defined.
Confidentiality
Even if you’re using ChatGPT as a translation tool, you’re unlikely to be allowed to. This type of technology cannot be used by companies for whom data security and privacy are of paramount importance.
Deployment options
Models like GPT cannot be used safely. Indeed, they are not separated from other users. There is only one ChatGPT model and it is shared between customers.
In conclusion, although ChatGPT seems to be an interesting low-cost translation tool, it is not appropriate to use within a company setting. Indeed, the low quality of the generated translations, the Google SEO algorithmic penalties, and the legal risks incurred are all negative points that should steer you away from this solution.
Want to go a step further?
