16 thoughts on “The era of blind faith in big data must end”

  1. Wow…Awesome ted talk about marketing.

    You really have the best collection of documentaries about marketing and advertising.

    Thanks for curating and sharing it.

  2. So, what this does, is reinforce what has always been said “those that have, get-those that don’t have, don’t get.” Which, very effectively kills the idea, of what America is or was about, allegedly, for those yearning to be free, free to peruse the happiness, prosperity; pursuit of justice, to live as humanly as possible. The “end-user” is being chosen by way of “written-code” written-off. And we all know ‘who’ the “end-users” are. Disconcerting, really, this ‘bias,’ what, incident or accident of technology? It has to be written in, right, or over-looked, realized, then ignored, perhaps?

    1. Have you checked marketing documentaries? I have curated some of the best documentaries about marketing and advertising. Do check out.

  3. Not sure if it is blind faith or is it just starting of the new era. Everything looks diplomatic when it is evolving from theory to practical use

  4. The algorithms are not the problem, but the people. By other hand, automated solutions executed by machines(computers) and only what humans do program it to do. The presenters blames the algorithm as if were something to blame, hahaha. People cheat, steal, lie, or is honest, and has nothing to do with algorithms.. yo are a bit dramatic, you should relax and re-elaborate you theory.

  5. I’m concerned ahout case of the two drug dealer. What if the formula prediction is true, which means the left has a crimnal risk of 10/10 while the right one, 3/10. Can we just sentence them according to this while they did quite same in the past? It’s unfair, right?

  6. It looks like the examples she used are algorithms that are created to be in a sense bias. Could we just solve this problem by not putting irrelevant information like race, gender, sexual orientation into algorithm and focus it on the unbiased facts?

  7. Here’s the ProPublica article that she was referencing: https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing . Unfortunately, the article doesn’t specifically say why those two defendants got their scores. (Perhaps because the company didn’t release full information on how they reached those scores.) The article does give examples from the survey used to run the algorithm:

    >> Race is not one of the questions. The survey asks defendants such things as: “Was one of your parents ever sent to jail or prison?” “How many of your friends/acquaintances are taking drugs illegally?” and “How often did you get in fights while at school?” The questionnaire also asks people to agree or disagree with statements such as “A hungry person has a right to steal” and “If people make me angry or lose my temper, I can be dangerous.”

    So for example, you might see higher scores for black criminals if black people are more like to have parents who went to prison, because of racism that existed in enforcement before the algorithms were implemented.

  8. I don’t think there is blind faith in big data (by those who practice it…).

    Big Data courses focus on measures of accuracy (of which there are several), on over and under fitting data (how well the model represents the data). They don’t say “HERE IS THIS PERFECT THING”, if they did there would be nothing for us to learn.

    I’m taking an AI course at the moment and I have recently trained a neural network to identify clothing.
    It passes the tests, it’s ~90% accurate.
    But I would never trust it with anything important.

    90% is very low.
    The data I used was not representative of the clothing that people wear.
    There is little information on what kinds of clothing have been missed and I haven’t checked that each item of clothing was represented at the same ratio as the clothing is likely to appear in the real world.

    All of these things are the kinds of things that real engineers have to think about in business.
    And they do.

    I’ve spent a little bit of time at some companies that work with big data and the engineers do have to think about their accuracy and validation.
    You can’t demo a feature without showing how accurate it is, how well it works over multiple different problems.

  9. Insightful talk….but my concern is that things are now swinging too far the other direction. That when the hard data is telling us something that we don’t want to hear, the politically correct data scientist is going to go in and wipe or distort the data to make it more palatable. For example, we all know there are real difference between men and women but we are not allowed to report these differences…..unless it is affirming that women are better than men at something. Same thing goes for race.

  10. I am sure this post has touched all the internet users, its really
    really fastidious paragraph on building up new website.
    I will immediately grasp your rss as I can not to find your email subscription hyperlink or e-newsletter service.
    Do you have any? Kindly allow me understand in order that I may
    just subscribe. Thanks. Ahaa, its nice discussion about this paragraph here
    at this website, I have read all that, so now me also commenting here.


Leave a Reply