Ariella Brown

Algorithms' Dark Side: Embedding Bias into Code

NO RATINGS
View Comments: Newest First | Oldest First | Threaded View
Page 1 / 5   >   >>
Lyndon_Henry
User Rank
Blogger
Re: Biased assumptions in algorithms
Lyndon_Henry   3/1/2017 3:25:54 PM
NO RATINGS
..

Ariella asks

Do you think that assumption may have been due to the bias the designers had about their own travel habits?



 

That was (and possibly still is) likely one factor. But the social context of the time probably also was a major factor. In the 1970s, the USA was just creeping out of the Transit Holocaust era (when the country's urban and regional transit infrastructure had been almost totally devastated, mainly by official policy and fiat) and traditional neighborhoods and historical structures were being systematically destroyed (something that Jane Jacobs campaigned against in her groundbreaking book The Death and Life of Great American Cities).

I'm nervous that a very similar "context" of adoration of a new transportation invention (robot cars) may be creeping back over the country, with associated impacts for the structure and habitability of our cities.

..

Ariella
User Rank
Prospector
Re: Just heard about the book
Ariella   3/1/2017 10:02:03 AM
NO RATINGS
@Seth @Michelle, Though the bias inherent in survey wording is a bit different than the problems of algoirthms, it is also something people should be aware of. I foudn this article on it .supersimplesurvey.com/blog/post/7-Tips-To-Minimize-Response-Bias

Michelle
User Rank
Data Doctor
Re: Just heard about the book
Michelle   2/28/2017 10:17:27 PM
NO RATINGS
@Seth I thought of survey design too. It does seem to a more difficult task to take bias out in this context, however.

SethBreedlove
User Rank
Data Doctor
Re: Just heard about the book
SethBreedlove   2/28/2017 7:05:37 PM
NO RATINGS
@ kq4ym  - That's a very good suggestion.  I've taken survery design courses that teach how to ask questions that are not biased.   I would imagine it would be harder to do in coding though because in surveys you are looking for information while in coding usually you are aiming for a specific result.

Ariella
User Rank
Prospector
Re: Just heard about the book
Ariella   2/28/2017 5:57:19 PM
NO RATINGS
@rbaz Indeed. Marketers always aim at particular targets, identifying what traits go into the populations that may buy what they're selling.

kq4ym
User Rank
Data Doctor
Re: Just heard about the book
kq4ym   2/28/2017 3:50:02 PM
NO RATINGS
For some time I've advocated consulting with sociologists, psychologists, ethicists and others who may have some valid input into how we propose our questions, analyze our data, and implement policies that will better lead to solutions that are non-discriminatory and valid using the science of behavior that we have at the time. Bringing various views into the mix can certainly lead to more rational decision making and maybe be less prone to errors.

rbaz
User Rank
Data Doctor
Re: Just heard about the book
rbaz   2/28/2017 11:59:16 AM
NO RATINGS
@Ariella, you are absolutely right. Targeting consumers based on race or ethnicity has always been done and is effective. Nothing inherently wrong with that, just denial of service or steering away intentionally is concerning and wrong.

Ariella
User Rank
Prospector
Re: Just heard about the book
Ariella   2/28/2017 9:23:44 AM
NO RATINGS
@impact While that sound fairly straightforward, I'd think working it out would be quite a complicated affair.

Ariella
User Rank
Prospector
Re: Biased assumptions in algorithms
Ariella   2/28/2017 9:23:00 AM
NO RATINGS
<The original models I scrutinized tended to generate more trips, more travel, but the algorithms were designed on assumptions that every household was in a low-density neighborhood (inner-city or suburban) and would be using cars for almost all trips. Also, the modal-split algorithms (deploying logit functions) seemed to incorporate the assumption that, for almost every traveler, you would have to pry their car from their cold, dead hands before they would ever consider boarding a bus or train.

So the planning models did predict huge travel growth, but almost entirely by personal motor vehicle, thus generating huge traffic volumes and implying the need for more and more roadway capacity (which, when actually constructed, then encouraged more traffic, thus fulfilling the models' predictions).>

@Lyndon_Henry I see. Do you think that assumption may have been due to the bias the designers had about their own travel habits?

impactnow
User Rank
Blogger
Re: Just heard about the book
impactnow   2/27/2017 10:33:18 PM
NO RATINGS
When the math is found to be discriminatory we need to address the issue by modeling to address the bias. While it seems like manipulation its more of a correction.

Page 1 / 5   >   >>
Information Resources
More Blogs from Ariella Brown
Social media and mobile devices may seem like a vast and scary unknown to parents of young children with phones and other devices. Now machine learning is being applied to the problem of protecting the kids.
Supply is not meeting the demand for analytics and data science professionals. Here's how one boot camp is looking to fill that gap.
Organizations around the world still have plenty of work ahead of them in terms of building trust in their data and analytics strategies, according to a report by KPMG.
Led by actor Gary Sinise, a pair of organizations are leveraging technology to provide smart homes to disabled veterans.
Radio Show
A2 Conversations
UPCOMING
James M. Connolly
Evolution of the Data Scientist Role


3/23/2017   REGISTER   0
UPCOMING
Jessica Davis
Monetize Your Data: Turning Insights Into Action


3/29/2017   REGISTER   0
ARCHIVE
Jessica Davis
Data Analysts in Training: Meeting Tomorrow's Demand


3/8/2017  LISTEN   63
ARCHIVE
Jessica Davis
Our Bodies, Our Data: Medical Records For Sale


2/21/2017  LISTEN   63
ARCHIVE
Jessica Davis
Energy Analytics: Using Data to Find Savings


2/14/2017  LISTEN   44
ARCHIVE
Jessica Davis
Sharpen Your Analytics & Data Management Strategy


2/8/2017  LISTEN   74
ARCHIVE
Jessica Davis
Analytics: Make the Most of Data's Potential in 2017


1/19/2017  LISTEN   19
ARCHIVE
Jessica Davis
A2 Radio: Can You Trust Your Data?


12/20/2016  LISTEN   70
ARCHIVE
James M. Connolly
Retail Analytics: See Where Style Meets Statistics


12/6/2016  LISTEN   53
ARCHIVE
James M. Connolly
Why the IoT Matters to Your Business


11/29/2016  LISTEN   45
ARCHIVE
James M. Connolly
Will Data and Humans Become Friends in 2017?


11/22/2016  LISTEN   40
ARCHIVE
James M. Connolly
We Can Build Smarter Cities


10/20/2016  LISTEN   31
Information Resources
Quick Poll
Quick Poll
About Us  |  Contact Us  |  Help  |  Register  |  Twitter  |  Facebook  |  RSS