Inside DeepL: How the startup is rivalling machine translation giants Автор темы: philgoddard
| | neilmac Испания Local time: 06:34 испанский => английский + ...
I tried Deepl out and it does quite a good job on the kind of texts I usually have to translate. The licence is also relatively affordable.
What I'd like to know however, is at what point does MT (machine translation) software become AI (artificial intelligence)? | | | Zea_Mays Италия Local time: 06:34 английский => немецкий + ...
AI is just the technology behind a tool and supposedly emulates human intelligence (as far as I understand it, it just does complex probability calculations - given "A", it calculates what most likely will follow "A").
So ChatGPT is not AI, it is an AI powered generative and dialog oriented chatbot. DeepL is an AI powered MT tool.
[Bearbeitet am 2024-01-26 14:18 GMT] | | | Peter Motte Бельгия Local time: 06:34 Член ProZ.com c 2009 английский => голландский + ...
Actually, there isn't a lot of intelligence in AI, because it uses statistics.
If in the texts with which AI is fed to be trained, "black" appears often in the neighbourhood of "stupid", AI will generate texts in which it calls blacks stupid.
And as a matter of fact, things like that often happen. It's one of the reasons that the plug had been pulled on an AI tool on Twitter once.
AI has nothing to do with intelligence, it only looks intelligent.
ChatGPT is astonishing b... See more Actually, there isn't a lot of intelligence in AI, because it uses statistics.
If in the texts with which AI is fed to be trained, "black" appears often in the neighbourhood of "stupid", AI will generate texts in which it calls blacks stupid.
And as a matter of fact, things like that often happen. It's one of the reasons that the plug had been pulled on an AI tool on Twitter once.
AI has nothing to do with intelligence, it only looks intelligent.
ChatGPT is astonishing because of the speed with which it produces grammatical, orthographically and structurarely flawless text, but there is no actuall intelligence behind it. ▲ Collapse | |
|
|
Is it really as good as claimed? I don't think so | Jan 27 |
I can't shake a suspicion that the claims of DeepL being by far the best MT engine are merely a well-organised PR campaign. Or maybe it really is better, but only for relatively simple texts like news articles. Last year, I compared DeepL and Google Translation on a couple of texts. One was a multilateral property acquisition and financing agreement, the other was a requirement specification for electronic hardware. I took a few excerpts from each and sanitised them to remove identifying informa... See more I can't shake a suspicion that the claims of DeepL being by far the best MT engine are merely a well-organised PR campaign. Or maybe it really is better, but only for relatively simple texts like news articles. Last year, I compared DeepL and Google Translation on a couple of texts. One was a multilateral property acquisition and financing agreement, the other was a requirement specification for electronic hardware. I took a few excerpts from each and sanitised them to remove identifying information but keep the phrase structure unchanged. Both MT engines failed to save my working time compared to unassisted translation, but DeepL was noticeably worse: the output flowed smoother, but many small yet meaningful fragments of the source text were silently omitted or badly distorted in the translation. One frequent type of error was the topic/comment swap, fixing which obviously requires the entire sentence to be rewritten. Even with the recent advances in MT, in my market segment (quality-conscious clients willing to pay high prices), none of the general-purpose MT engines seems to cut the mustard, and the only custom engine I know that produces a workable output is not a pure MT but a hybrid of MT with a large TM compiled by a big agency. ▲ Collapse | | | Arne Krueger Германия Local time: 06:34 немецкий => английский + ...
...in a 2023 Mercedes car.
Issue: Rear door locked itself.
Using AI-powered info system...
"Hey Mercedes, how can I unlock the rear door?"
"You can adjust your door mirrors by using the comfort button on your dashboard."
After my fourth attempt, I gave up and climbed out of the other passenger door... | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Inside DeepL: How the startup is rivalling machine translation giants Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
| Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |