Hi there,
I’m curious to pick your brains on an obscure subject.
Is AI for federal purposes going to displace linguists from many roles, as the NLP algorithms for these air-gapped systems become ever more sophisticated in more languages?
There is now a Chat GPT version for federal purposes, as well as Langley’s own buildout of indigenous AI systems for foreign language work, so I’m curious whether anyone is seeing a dwindling of cleared foreign language work in the space.
There’s a decrease in private sector translation for commercial purposes due to the enormous improvements in the algorithms, but will it be the same for cleared work in, say, the IC?
There isn’t much public information due to the classified nature of such information.
This is probably a more appropriate topic to discuss elsewhere, as I believe revealing information like this may not be in the best interest to discuss publicly.
About 25 years ago I worked at a company that was developing a human language translation capability. It required a great deal of “training,” using sample data to train the model; also, it could only handle certain limited subject areas. It was quite a breakthrough for its day.
The technology available today is so far beyond that it is dizzying.
One key point is the need for a human to perform the translation, even if they are using these tools, and another human to review the translation. That will keep some people employed, although probably not as many as in the past.
What will be the next stage? AI ‘bots’ plowing through raw data generating all source reports? Given the ever growing amounts of data that need to be reviewed, it will surely be a temptation to make the human role a last sanity check. Yes, today there are some rather rough translations coming out of tools like Google Translate (and whatever the IC equivalent might be) but they’re getting better and better.
At any rate, I think the days of someone making a career out of being a language analyst are gone. The exciting work lies elsewhere… as do the promotions, no doubt.
this can only work to an extent. AI cannot read inflections or dialects.
Yes it can, more and more every day. Just keep throwing data at it 