Working On Microsoft's Cortana Is Laborious and Poorly Paid (vice.com) 19
An anonymous reader quotes a report from Motherboard: Apple, Google, Amazon, and most recently Facebook have been found hiring human workers to transcribe audio captured by their own products. Motherboard found Microsoft does the same for some Skype calls, and is still doing so despite other companies suspending their reliance on contractors. A cache of leaked documents obtained by Motherboard gives insight into what the human contractors behind the development of tech giants' artificial intelligence services are actually doing: laborious, repetitive tasks that are designed to improve the automated interpretation of human speech. This means tasks tech giants have promised are completed by virtual assistants and artificial intelligence are trained by the monotonous work of people.
The work is magnified by the large footprint of speech recognition tools: Microsoft's Cortana product, similar to Apple's Siri, is implemented in Windows 10 machines and Xbox One consoles, and is also available as on iOS, Android, and smart speakers. The instruction manuals on classifying this sort of data go on for hundreds of pages, with a dizzying number of options for contractors to follow to classify data, or punctuation style guides they're told to follow. The contractor said they are expected to work on around 200 pieces of data an hour, and noted they've heard personal and sensitive information in Cortana recordings. A document obtained by Motherboard corroborates that for some work contractors need to complete at least 200 tasks an hour. The pay for this work varies. One contract obtained by Motherboard shows pay at $12 an hour, with the possibility of contractors being able to reach $13 an hour as a bonus. A contract for a different task shows $14 an hour, with a potential bonus of $15 an hour. A Microsoft spokesperson told Motherboard in an emailed statement, "We're always looking to improve transparency and help customers make more informed choices. Our disclosures have been clear that we use customer content from Cortana and Skype Translator to improve these products, we engage third party expertise to assist in this process, and we take steps to de-identify this content to protect people's privacy."
The work is magnified by the large footprint of speech recognition tools: Microsoft's Cortana product, similar to Apple's Siri, is implemented in Windows 10 machines and Xbox One consoles, and is also available as on iOS, Android, and smart speakers. The instruction manuals on classifying this sort of data go on for hundreds of pages, with a dizzying number of options for contractors to follow to classify data, or punctuation style guides they're told to follow. The contractor said they are expected to work on around 200 pieces of data an hour, and noted they've heard personal and sensitive information in Cortana recordings. A document obtained by Motherboard corroborates that for some work contractors need to complete at least 200 tasks an hour. The pay for this work varies. One contract obtained by Motherboard shows pay at $12 an hour, with the possibility of contractors being able to reach $13 an hour as a bonus. A contract for a different task shows $14 an hour, with a potential bonus of $15 an hour. A Microsoft spokesperson told Motherboard in an emailed statement, "We're always looking to improve transparency and help customers make more informed choices. Our disclosures have been clear that we use customer content from Cortana and Skype Translator to improve these products, we engage third party expertise to assist in this process, and we take steps to de-identify this content to protect people's privacy."
So? (Score:1)
Why ? (Score:1)
Said the whole world: (Score:2)
Excuse me, what's Cortana?
No time for personal data (Score:2)
If you're trying to slam out work that fast then even if you hear some personal stuff it's not like you have time to compile your own set of notes.
Shirley, you can understand the allure (Score:5, Funny)
Low pay and long hours, now; But: ...dying in your bed many years from now, would you be willing to trade all the days from this day to that for one chance, just one chance, to be proudly boasting that you were working on the Cortana team?
Re: (Score:2)
Enough already (Score:5, Informative)
All supervised machine learning systems require labeled training data. All production speech recognition systems are included in this class. There is literally no way to do this without human annotators. All companies building speech recognition systems, by definition, will be transcribing sampled data. Anybody who is surprised by this does not understand what machine learning is and thinks it is some kind of magic. It is not. Nothing Google, Amazon, Apple or Microsoft is doing here is wrong, surprising or undisclosed. Stop panicking about it - if you don't like it, don't use speech recognition systems.
Re: (Score:3)
Also, since when is a typical job not "laborious"? Love the click-bait / manufactured outrage here.
Sure (Score:1)
Re: (Score:2)
Joe Sixpacks, i.e., the bulk of users, have no idea how voice recognition systems work nor have likely even heard or machine learning. The fuss is largely about the fact that the companies didn't disclose that humans would be eavesdropping.
Re: (Score:2)
There is no doubt speech recognition system requires human annotators for training.
How they get those samples is another matter. ...
You could for example use TV/Movie captions as a trainer.
You could ask people to clearly opt in or clearly let them know they are being recorded as if you called a help line and it says 'this call is being recorded for quality..."
It's not the fact that they're annotating it. It's how they are getting the source data.
As to the other bit about the pay being poor. This is fundamen
People doing boring repetitive tasks? (Score:3)
Hasn't this been a thing since forever? People being paid poorly to do trivial, repetitive tasks?
Technology and AI will eventually eliminate this kind of work. In the mean time, you can expect when using something like Cortana or Siri that it is highly likely a human could hear some of the voice data you've submitted to the system. To think otherwise is totally naive .
Plus (Score:2)
Preposterous (Score:1)
Quote: "we take steps to de-identify this content to protect people's privacy."
Translation: "We don't know what's in the audio, but we lie to you we do some "steps" to de-identify the content"
Example: the audio says "Michael, next week, let's go to the sexshop at Liberty Street and buy a dildo for your brother. Don't forget that we will celebrate at John's Big Restaurant in Munchin City"
Now, if the "human" translator is from KKK or Radical Muslim or Radical Christian or etc. there they have private informat
BS (Score:1)
GNU Messenger (Score:1)
Sad, really (Score:2)
If one thinks of all the money those Cortana users bring.
Both of them.
Great! Low-skilled jobs! (Score:1)