Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Bard Is Trained by Humans Who Say They’re Overworked, Underpaid and Frustrated (bloomberg.com)
46 points by vector_spaces on July 12, 2023 | hide | past | favorite | 7 comments


Sounds like Google's translation team complaints.

> Ensuring that the response is well-sourced and based on evidence, however, falls to thousands of outside contractors from companies including Appen Ltd. and Accenture Plc, who can make as little as $14 an hour and labor with minimal training under frenzied deadlines, according to several contractors, who declined to be named


> Even if the inaccuracy seems small, “it is still troubling that the chatbot is getting main facts wrong,” said Alex Hanna, the director of research at the Distributed AI Research Institute and a former Google AI ethicist.

it didn’t “get the main facts wrong.”

you can’t blame the model for the failings of the modelers.

this stopped being pedantic when monkeys decided to hoover the internet, digest it, and throw it at the wall to see what would stick.


At least it's trained by humans. In 50 years Bard will train humans to be more efficient in its neo-corporate culture, where Bard is the CEO.


That’s silly. There’s no free lunch in the generative AIs. Emergence from The Pile is a myth. You get what you put in.



Google failed so much on the AI, this company R&D is a joke.


Bard isn't very useful now relative to ChatGPT, but I still have little doubt they'll have the main product in this area eventually.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: