The current discussion about AI is the stuff of nightmares for many of us that practise the craft of texts … and then the waking soul comes across the 50,000-like post explaining that AI destroyed the translation industry years ago, before, errr, getting back to work.

I’ve been repairing the output of one of these products recently. I have caught it anointing somebody to the priesthood at random, describing a figure suspected of sexual adventures as ‘celibate’, talking about a couple when one partner in the relationship was actually being described. There is more.

One should not overreact. I have used such tools in my own research on occasion, to help get the gist of secondary literature in a language in which I’m not fluent. It is astonishing what they are capable of, in a good way. But for the specialist texts I work with, they are simply not adequate. There is, for the foreseeable future, no getting around the need for a real person.

Besides, humans do more than just fix what the robots get wrong. In this case I’ve also been able to suggest little tweaks that make the structure of the argument easier to follow, and picked up a few cases where material had crept into plot summaries that followed the primary text verbatim without being marked as a quotation.

The commercial point in all of this is that the client appreciates the challenges and pays a reasonable rate that covers the time, expertise, research, and experience necessary to address them. And they know it still won’t be quite as good as if I had translated it all by hand.

As others have pointed out, the issue isn’t so much the machines per se, as how humans work with them – if people are taken in by the marketing and think that, if a safety net is needed at all, a quick proofread is enough. Or if organizations and agencies know that the output is flawed but pretend it isn’t in order to pay the humans who vet it a pittance. The results of that are predictable.

Like so much else, it comes back round to personal and long-term relationships with clients, creating a business space outside the race to the bottom when it comes to pricing and expectations.