zlacker

[parent] [thread] 13 comments
1. KirinD+(OP)[view] [source] 2019-07-16 15:27:16
> What did the "Open Research Group" at Google actually build?

Have you made any effort to investigate who Meredith Whittaker is on your own?

Her work on AI ethics was much appreciated and celebrated precisely because she was a distinguished contributor. The cultural aversion to building weapons is not novel thing in that culture.

replies(3): >>gdy+re >>Merril+dk >>bougal+T01
2. gdy+re[view] [source] 2019-07-16 17:06:23
>>KirinD+(OP)
"Her work on AI ethics"

What exactly was that?

replies(1): >>depart+4h
◧◩
3. depart+4h[view] [source] [discussion] 2019-07-16 17:24:17
>>gdy+re
Do you not know how to use google?

<https://ainowinstitute.org/research.html>

replies(2): >>gdy+0s >>dang+Rp1
4. Merril+dk[view] [source] 2019-07-16 17:41:56
>>KirinD+(OP)
Yes, I googled Meredith Whittaker before commenting. She joined Google in 2006 after getting a bachelor's degree from Berkeley, apparently in literature and psychoanalytic theory. By 2012 she appears to have been a program manager in initiatives regarding internet measurements. So far I haven't found any references to products or reports from the Open Research Group, which she appears to have founded sometime prior to the 2016 White House conference where she met the other co-founder of the AI Now Institute at NYU.

Typically when you see someone engaged in "technology ethics" their professional career is based on limiting or stopping the technology, rather than building or advancing the technology. See, for example, stem cell ethics. Companies don't typically set up adversarial organizations within themselves. A more usual approach is to set up temporary "red teams" to address specific issues.

replies(1): >>skybri+5o
◧◩
5. skybri+5o[view] [source] [discussion] 2019-07-16 18:09:51
>>Merril+dk
I only know a little about it, but it seems like financial companies commonly have a permanent compliance department?

https://www.investopedia.com/terms/c/compliancedepartment.as...

replies(1): >>Merril+su
◧◩◪
6. gdy+0s[view] [source] [discussion] 2019-07-16 18:37:37
>>depart+4h
You missed the word 'exactly'. Can you summarize her work so that it doesn't sound trivial or bogus?

Just looking at the titles I expect something similar in quality to articles debunked here: https://arxiv.org/abs/1905.09866

replies(2): >>KirinD+Ay >>xyzzyz+8E
◧◩◪
7. Merril+su[view] [source] [discussion] 2019-07-16 18:54:39
>>skybri+5o
This is true, but the function of the compliance department is to ensure that the company operates within the boundaries set by law and regulation. They typically operate with the authority and access to top management. They would report incidents of non-compliance to management and much more rarely to a regulator, very rarely to the media. If members of the compliance department report to the regulators or public they are covered by the relevant whistleblower statutes. I can't recall members of the compliance department organizing other employees to demonstrate.

Another organization is the quality control department of a manufacturer. They also tend to report independently to top management, and they function similarly.

◧◩◪◨
8. KirinD+Ay[view] [source] [discussion] 2019-07-16 19:30:29
>>gdy+0s
So is the insinuation here that if she doesn't meet some arbitrary series of qualifications you put forth, it's okay for her to be illegally harassed out of her job?

I'm reluctant to engage further because it seems like an absurd line of reasoning.

replies(1): >>gdy+iE
◧◩◪◨
9. xyzzyz+8E[view] [source] [discussion] 2019-07-16 20:11:03
>>gdy+0s
I clicked on the first one with her name, and the main conclusion was that AI needs more workplace diversity.

I clicked on the second one with her name, and the main conclusion was that AI needs more government regulation, labor unionization, and yes, you guessed correctly, workplace diversity.

I begin to feel that AI is only a red herring here.

◧◩◪◨⬒
10. gdy+iE[view] [source] [discussion] 2019-07-16 20:12:50
>>KirinD+Ay
Nope, you've made yourself a straw man. I hope you did not do it intentionally because it's plainly disgusting. Please don't do that again.

You've said that she didn't just leave Google instead of protesting because she didn't 'want to see the thing you [she] worked so hard to build misused to build killer robots and "war minds"'.

You was asked what she actually did at Google and you've come up with 'her work on AI ethics was much appreciated and celebrated' as a response.

Looks like she was not working 'so hard' on anything that can be of any use for building 'killer robots and war "minds"'. In fact, for building anything.

replies(2): >>KirinD+jG >>dang+Wp1
◧◩◪◨⬒⬓
11. KirinD+jG[view] [source] [discussion] 2019-07-16 20:30:53
>>gdy+iE
> Looks like she was not working 'so hard' on anything that can be of any use for building 'killer robots and war "minds"'. In fact, for building anything.

This looks like you've subtly dodged the complaint I raised by trying to insinuate that she doesn't have a right to express opinions aboht functions of Google she herself did not personally participate in. This leverages the information asymmetry in disclosure; we can't publicly discuss the bulk of her work and therefore you can suggest that there was none.

I find this to be no different from suggesting that she has no right to protest and therefore deserves to be run out. You're just trying to run the standard "she wasn't that important and therefore doesn't have credibility" playbook. Gross.

But despite the disingenuous argument, I'll accept it head on. I challenge the entire premise. I certainly can and do express opinions about my employer's involvement in weapons development and I am glad they are not doing it. I'd fight to avoid doing any more of it, and I'd be willing to resign over it. I don't work in AI, but my work supports any such system at Google and therefore I'd feel responsible to help prevent building killer robots in any capacity.

I, like many such employees, am both a shareholder and an employee in a company that claims to have an interest in a transparent and egalitarian corporate culture. This practice will naturally introduce friction between different parties and I expect us to work through them as fairly as possible. So I will not simply eject at the first sign of something I don't like. But if I feel that there is a line crossed while I was there strenuously objecting to that line and I haven't been given adequate reason to change my mind, I won't hesitate to resign.

I make these facts clear to folks when they hire me. If they don't like it, they shouldn't hire me. Hopefully you respect your own agency and intellect enough to give yourself similar license in your own life.

12. bougal+T01[view] [source] 2019-07-16 23:25:13
>>KirinD+(OP)
There is no such thing as the 'Open Research Group', outside of Meredith. see https://googlersagainstdeceit.blogspot.com.
◧◩◪
13. dang+Rp1[view] [source] [discussion] 2019-07-17 04:16:01
>>depart+4h
Please don't cross into personal attack and snark, regardless of how wrong another comment is or you feel it is. That only makes this place worse. Your comment would be just fine without the first sentence.

https://news.ycombinator.com/newsguidelines.html

◧◩◪◨⬒⬓
14. dang+Wp1[view] [source] [discussion] 2019-07-17 04:17:18
>>gdy+iE
You crossed into personal attack and name-calling in this thread. Would you please review the site guidelines and follow them when posting here? Note particularly: "Comments should get more thoughtful and substantive, not less, as a topic gets more divisive."

https://news.ycombinator.com/newsguidelines.html

[go to top]