Machined Prejudice

Machined Prejudice

Reading Time: 1 minute

Machined Prejudice

Technology bias can amplify the worst of human nature. How do we keep it out of artificial intelligence?

We’re now seeing more and more cases of institutional and individual human bias creeping into search engines, social media and other technological systems. How this bias infiltrates our tools in is an old story, of course, but we’re starting to see new types of bias that say a lot about the way we build and use artificial intelligence. 

Technology bias is a problem that few of us think much about, but which has the potential to be incredibly important as more and more of our economy — and society — come under the control of artificially intelligent systems. 

#technology #bias #MachineLearning #culture

http://www.the-vital-edge.com/technology-bias/

42 comments

  1. IMO this has a lot to do with the statistical nature of most machine learning algorithms.

    Case in point: Microsoft’s Tay bot. So much cursing and l33t talk feeded into it inevitably led to it using the same language.

    IMO a mix of an expert pre-coded formal education, plus machine learning, could be a partial way of handling this problem.

  2. Agreed, Dimi 7. In some ways, the training process is going to have to look a lot more like selecting statistical samples for research purposes and ensuring that we don’t introduce bias into the sample set.

    I think your point about pre-coded, formal education is an interesting one. Might we be able to automate the pre-school years through some basic training to stabilize certain factors? Not sure exactly what that would look like, or how generalizable that is, but it’s an interesting idea.

  3. Gideon Rosenblatt Neural nets are very vulnerable to biased data in the early training phases. That’s why I’m proposing a formal education which must be given a hardcoded bigger weight in the neural net.

    Of course, AI is a wild west area still, but it’s a good start IMO. There are people who code expert systems. We can learn a lot from them as well.

  4. Yeah, I’ve got a feeling that we’re going to see a lot more melding of techniques in the years ahead, Dimi 7. Did you read Pedro Domingos’ book?

  5. Gideon Rosenblatt I’ve only recently started getting some time for my own activities, so nope. But I’m strongly interested. Want to send me a link?

  6. Gideon Rosenblatt Thanks! I’ll also read some Kevin Kelly, too.

  7. keep the accountants away from AI and instead use poets and musicians to train them.

  8. Gideon Rosenblatt I mentioned before off-the-mark personalized search results. Bias is. I find it frustrating.

  9. Any particular pattern you detect, Zara Altair​?

  10. Gideon Rosenblatt Yes, information searches vs. kebabs near me. On Google search I receive what Google thinks I want which oftentimes is way off the mark, especially research in history. Results are better if I go to google.fr or google.it…so far. I’ve tried going Incognito but that not only doesn’t help, the results are worse.

    For instance in the U.S. if I enter a search query in another language the results shown are in English. I know my search queries are quite individualistic but I find the results worse than before “personalized.”

  11. That’s weird, Zara Altair​. I can’t understand why that would be the case. In other words, in terms of thinking of this as ‘bias’, I don’t understand how a worse experience is created by including the additional personal information. I’m not doubting it, just trying to see how it might happen.

  12. Gideon Rosenblatt I don’t know. Perhaps because Google sees my first language as English and screens out the more pertinent information in other languages. It used to be the case that I automatically had search results in other languages, especially if they were pertinent to my query.

    Kebabs near me works fine. 🙂

  13. Gideon Rosenblatt once I’ve done a generic (but obscure) search I turn to Google Scholar.

  14. Gerallt Franke True to the last word.

    Not to mention that if we teach AI to draw knowledge from us and respect what has been achieved before its conception, that might help in it not ending up wanting to kill us, like Skynet.

  15. I can speak from first hand knowledge on being subjected to this phenomenon.I can also map some of it out for you.

  16. Fascinating peek into how inherent biases affect everything, Gideon Rosenblatt. First, how do we address in “real” life & then in AI.

  17. Thanks Nadine Hack​.

  18. I’d be interested in hearing about it, Anthony Jones​.

  19. On the AI matter i have suggested a study that uses the Bible as the sole input for an ai or predictive analysis model to determine if man’s way is not the way of God.

  20. Gideon Rosenblatt​ I’ve always favored virtual intelligence over artificial intelligence. If the machine is not self programming I’m not sure that we have an issue with bias at that point.

  21. Anthony Jones can you expound on this. it sounds interesting.

  22. You have to work with the mental challenges and physically debilitating half now in front of me and make no mistake about it Gideon I love reading your blog. I will be right back.

  23. The Future Of A.I.

    British scientists have just been given the go-ahead to genetically modify human embryos. There are grave concerns, however, that this latest development opens the door to ‘designer’ or GM babies. Gene editing, which enables DNA to be modified, is just another step towards the ultimate goal of genetic engineering: to create an artificially intelligent human being. So far the results have been disappointing but nothing can stop what’s coming in the pursuit of the ultimate scientific dream.

    The difficulty for scientists has been in replicating the creative function of the brain, which will enable the robot to respond with the precision of human intelligence in any given situation. However, a scientific breakthrough is pending in which a remarkable biologically engineered gene will replicate the self-reflective intelligence of human beings. These human robots will be artificially self-conscious, animated through biological implants which seamlessly replicate, in appearance and action, the functions of man and woman. As amazing as they appear, the human robots will have a major flaw: they will be devoid of love and unable to be conscious of the source of their inward intelligence.

    The new technology will revolutionise the world in ways beyond our comprehension. The most obvious indication of global cultural change will be the ever-widening gap between the rich and poor. The national security services will deploy A.I. combat soldiers to police the growing civil unrest, which will result in residential areas for the wealthy and privileged classes becoming inaccessible to the masses.

    The subconscious impulse that motivates the scientific mind to produce such technological marvels has its roots in the primal mind-stuff of the immemorial past. Slavery, although no longer politically correct, is deeply embedded in the human psyche as a complex of intense desire patterns for control and power over others. The emotional craving to enslave has never gone away, only imperceptibly filtrated through the global scramble for comfort and convenience. The invention of A.I. human slaves will be seen to fill this niche. The primary role of these robots will be to function as sexual slaves for the pleasure of man to do anything he wants without needing to be responsible for his actions.

    Love between the genders will all but disappear. For what could compete with relinquishing any sexual restraint that inhibits man’s animalistic drive to dominate another? Woman, the female principle, will have no choice but to follow the example of man. She will inwardly despise his betrayal of her and go along with the virtual performance of sex without love. The pain of child birth, which woman willingly and valiantly undertakes on behalf of life on earth, will no longer be necessary. The transference of the conception of life from earth womb to sterile laboratory symbolises the externalising of the scientific rational mind. It’s not the technology that’s the problem but the degeneration of virtue that comes from abandoning love on earth for a manufactured replica of the reality behind the body. No good whatsoever will come from Artificial Intelligence, except as an evolutionary phase of human culture consonant with the deepening descent into human ignorance and the exploration of the objective universe.

  24. Lance, you have just written a lot about yourself. Then generalised that everyone is like you.

    Sorry but your above post whilst written well is nothing more than a dark male stereotypes version of what might happen with AI.

    Even if a man was really into dominating a sexual partner they would soon get bored of an AI as where is the fun in dominating something that can not feel properly. You take the meaning out of domination and thus your dark logic fails.

  25. Do I Detect Valid Bias and Assumptions here already..? Ok I’m going to Generalise MY Own Opinion and Agree with Lance Kelly and Set about My Dark Grave Disconcerting Parallels to that, of which, his inferred stated Logic has been mentioned. Unless, I can Eliminate my Seed of Doubt for the 100% Surety of AI as having to be an Impertinent attachment part of my Anatomy, I Strongly will not adhere or neither concede to its full Implementation unless otherwise proven its Safe-Proof Potential and Hazard-Free Warranted Worthiness.

  26. A machine will have the brain in the form of software to the extent of intelligence given by a human and hence it will be able to perform only logical functions and not emotional functions…

  27. Gopalbhai Trivedi ….That is ” EXACTLY WHAT I DO NOT WANT..Happen to have a HALF-AUTOMATED Blue-Chip inserted in My Neuro-Skull …Where it CAN DEFINITELY be ACCESSED and at a FLICK-of-a-Switch SOME FRANK-EINSTEIN Insane-in-the-Membrane Can PLAY-ME-for-a-PUPPET…,Taking over my Head-Space BRAIN..,.and TOTALLY

    MANIPULATING & CONTROLLING my body..,thru’ this FANTASTIC DEVELOPMENT ….No,..No,…No,…!!

  28. People need to get a grip about AI – at present and until we get some massive break through in science AI remains as good as Siri, Google Now and Cortana. In other words it is about as intelligent as my ass.

    Intelligence is freedom of thought and might even be part of a “soul” if you believe in this. A life force might be a better way of describing it. You can not just attach this to a machine. You can not create code that enables intelligence to exist, you can just mimic it.

    So if you want to sleep with a machine that is pretending to be a human then good for you but you would one odd cookie.

  29. Gopalbhai Trivedi ..So GO A HEAD and GET it DONE to Yours I say..,and GO SURF OUT in SPACE…because that’s WHERE this IMPLANTED FIASCO is LEADING too…” Out-of-MIND-and-OUT-of-SIGHT…NO DEAL until it’s PROVEN A REAL CLONED ASSET and REALLY TOTALLY NECESSARY ….Highly SUSPICIOUS FOREVER I AM…..

  30. I have a very impressive chatbot. But i wrote it in javascript so except running on a device while i show it, it will never be seen by anyone. It learns verbs based on nouns you teach it based on the key word dictionary i defined.

  31. I feel personally that as a planet we still have lots of growing up to do before its safe to even go down these lines. Far too many people still got a controlling mentality I’m afraid. I’m not against it I just know it will get used against me still at present. SIMPLE WE’RE NOT GROWN UP ENOUGH YET!!!!!! UNFORTUNATLEY

  32. I agree Leena – just read the post by Lance Kelly above….AI for sex purposes – grow up!

  33. lol, someone thinks they wont get it ever if they maake sex dolls? lol sex dolls have been around a long time, adding ai to one is a waste of time in the first place…. who talks?

  34. I am amused by the concern to keep “humanity” out of artificial intelligence. Folks, those horses are out of the barn and in the neighbors pasture. Most of you do no recognize them. A few names may help. Google. They regularly collect all kinds of information about you and sell it. Where is the AI part? It is hidden in several places, but their search engine is one of them. Another is Facebook. Where is the AI? Again, it is well hidden, but it is there connecting the dots, making predictions about who should see your posts and who should not. It is used in many other ways. FB is actively doing research on AI for many application areas in their application.

    Are these people/applications benign? Are they helpful to you as their owners want you to believe. No and no. Detailing is a long blog.

    Is there a solution to these problems? Definitely yes. I have created several models for how people interact in ways that are mutually beneficial or benign. I use the word, useful relationships. The solution to the problem is take these models and create artificial intelligence using these ideas. This application can test and probe other applications for adherence to the “useful” model. If an application shows evidence that it is not behaving according to the model for useful relationships, the person can be notified.. and many other approaches can be taken.

    If someone in the real world of human relationships fails to adhere to the model for useful relationships, I walk away. I have lots of friends. I don’t need any who show themselves “not useful.” As a practical matter, I have had to walk away from a few people in my life, but less than 10.

  35. Hello and a good day to You.

    Thank You for this post. I did not get finish reading on the Internett. .but I will be getting back to it. Very interrsting. Indeed. I recognize the “pattern” here.

    I have been seeing it on Google .. and its machineguns and Beautiful Flowers hand in hand .. He he.. its incredible. Thnx ! 🙂

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up here for the latest articles. You can opt out at any time.


Subscribe by email:



Or subscribe by RSS:

Subscribe-by-RSS---Orange-Background
%d bloggers like this: