Ever heard of “vegetative electron microscopy”?
Sounds impressive, right?
Well, it's nonsense... but that hasn't stopped it from being quoted in 22 scientific papers.
It's a “digital fossil” – an error preserved and reinforced in AI systems, in this case, resulting from a series of errors stemming from a mistranslation from Farsi (all because of one dot!).
This is a classic case of 'AI contamination' and a product of GIGO – garbage in, garbage out.
Experienced programmers, translators and academics have been warning about this for years for code, translation memories (TM) and now the large language models behind AI chatbots like ChatGPT.
For instance, to keep costs down, many language service providers have been building TMs using dubious input from cheaper and less qualified practitioners. The output of these TMs is then often unreliable and incoherent, i.e. garbage!
Once these errors enter these systems, not only is it near impossible to remove them, but they end up perpetuating and amplifying. It's best then to start from the beginning.
For more on the curious digital fossil "vegetative electron microscopy", check out this article here
I can give an example of a digital fossil for Serbian>English translation: "dr. zakon". It should be translated as "as amended" but stemming from an incorrect but upvoted translation on ProZ in 2008 and then fed into TMs and AI, it's far more commonly translated as "state law". The damage has been done.
But, if you want to make sure that your text is translated free of any damage or garbage, why not drop me a line at info@nicknasev.com and let's discuss!
.%20A%20day%20of%20campaigning%20%E2%99%80%20%E2%80%A6%20or%20a%20day%20to%20buy%20flowers%20%F0%9F%92%90.jpg)
