AI And “Hate Speech”

Thoughts on Zuckerberg’s thoughts, from a gamer.

Proposals to outlaw "hate speech" make me feel unsafe. https://t.co/bPc3ZPEbdK

— Rand Simberg (@Rand_Simberg) April 11, 2018

[Update a few minutes later]

This book, which every incoming college freshman should read (it’s not long) seems relevant.

27 comments on this post.
  1. Paul Milenkovic:

    Facebook should encourage hate speech so they can track the people who post it or even view it along with their entire network of “friends” and the friends of those friend.

    Fools!

  2. peterh:

    In a sane world, very little of what is described by the left as “Hate speech” is legally actionable. I see far more from the left that can reasonably be taken as a threat or advocacy of criminal action.

    What am I saying, “sane world”? In the leftist world their hate is all they need to justify brutal punitive action. And because “everything is shared”, their hate becomes “our hate”.

  3. Paul Milenkovic:

    It may not be legally actionable, but it is certainly socially actionable — think of Mr. Hogg’s fine boycott or think of Google firing someone responding to a call for an internal “conversation.”

    There are many things that can be done to you that don’t involve the law. Why would Facebook want to filter out hate speech — they should encourage it so correct thinking people can pick out all of the haters — and their “friends” and the friends of their friends.

  4. Edward M. Grant:

    ‘Hate speech’ is the ‘assault weapon’ of the First Amendment.

  5. Paul Milenkovic:

    I don’t agree with that analogy. There a lot of things that people say as “hate speech” that may not reflect hate as much as a desire to say provocative words to draw attention to themselves. You cannot say the “F” word an get much of “a rise out of people” whereas the “N” word can certainly do that. Why do people wear shirts with skull images on them? I don’t think they really belong to a death cult — they are just trying to be provocative.

    That said, are there not legitimate reasons — for shooting sport, for lawful home defense, for hunting, for protecting one’s self, family and home in the event of a civil disturbance, for deterring the emergence of a tyranny, both through widespread possession by the people as well as the people being trained in marksmanship applicable to personal arms or to arms that may be issued to the people — to possess and to learn how to operate a semi-automatic rifle or pistol?

    Are you saying that persons who own semi-automatic firearms are doing this to be provocative rather than quietly exercising their natural rights protected under the Constitution to possess these arms for the listed and other reasons?

  6. wodun:

    He is just saying hate speech is being used as a boogie man to justify taking away rights and restricting activity just as “assault weapons” are.

  7. Edward M. Grant:

    Yes. It’s a meaningless term that can be used to scare the uninformed into supporting any kind of ban.

    ‘You don’t support people owning assault weapons, do you?’ they say as they push a law to ban single-shot .22s. ‘You don’t support hate speech, do you?’ they say as they push a law to censor anything a man says.

  8. James:

    Rand,

    Mill’s original On Liberty is available as an Amazon Kindle edition for the astounding sum of $0 and 0¢.

  9. MarcusZ1967:

    On Liberty

    http://www.gutenberg.org/ebooks/34901

  10. peterh:

    “Hate speech” is speech they hate.

  11. wodun:

    And that is why it can’t be defined. It is really less about the speech and more about the people speaking.

    Clear definitions could lead to equal application of the rules but the rules are not meant to be equally applied. They are meant to target speakers, not speech.

  12. David Spain:

    AI to eliminate “Hate Speech”(tm)? Artificially of course.

  13. Bob-1:

    I don’t understand what is meant by “outlaw” in this context.

    I’m unconcerned about what Rand has on transterrestrial musings, nor am I concerned about what Mark Zuckerberg has on his little website project. It is really nice of Rand and Mark to let random people on the internet post comments to Rand & Mark’s respective websites, but I think it is to be expected that Rand and Mark will be the arbiters of what shows up.

    If Mark starts censoring what Rand can say on transterrestrial musings, then I’m concerned!

    Are you worried that Mark will donate his AI programs to internet carriers, and that the net will no longer be neutral? No, wait….

  14. McGehee:

    How about if Rand were to collect personal data on you from other users, and sell it? Even though you had never even visited his site? Would that concern you?

  15. Curt Thomson:

    Or how about you are admitted to the hospital for a procedure and Mark obtains information about you, such as what the medical condition was, what drugs you’ve been prescribed, what information you provided regarding your lifestyle. And he sold that information.

    Even though you had never even interacted with facebook in any way. Would that concern you?

  16. Leland:

    Or what things you are allergic to and made sure others knew?

  17. Bob-1:

    McGehee and Curt, the answer to all of your questions is yes I sure am concerned about all of that. I think free enterprise needs to be thoughtfully regulated so that it stays free but not quite free enough to behave like facebook.

    But the behaviors you’re asking about seem rather different to me from Zuckerberg’s plan to censor speech on his own website. I’ll alluded to this, but I’ll ask directly: Do you think Zuckerberg censoring speech on facebook is different from the rare times that Rand has very reluctantly censored comments he didn’t like on Transterrestrial Musings? Zuckerberg’s justification is certainly different from Rand’s, but do you think Zuckerberg isn’t well within his rights, just as Rand is, to censor whatever he wants?

  18. Jon:

    Rand has never censored you even though you have opposing political views. Facebook shuts down conservatives all the time.

    Since Facebook is a private company, they can do what they want. There are alternatives like minds.com. Facebook’s active users base is declining, from what I know. Gen Z looks at it as something grandma and grandpa use. Many just use it for the IM.

  19. Karl Hallowell:

    Since Facebook is a private company, they can do what they want.

    They still can’t commit fraud. Sure looks to me like Facebook promised one thing to its users when they were building up their user base and competing with rivals and now that they don’t have that competition, they’re promising a different thing. Could be a bait-and-switch here.

  20. wodun:

    I think free enterprise needs to be thoughtfully regulated so that it stays free but not quite free enough to behave like facebook.

    Our system of governance and society depend on a mostly virtuous people. Those virtues come from the long history of Western Civilization. Some of them are codified and some of them aren’t.

    The ones that aren’t codified are just as important as the ones that are. The problem is that our friends to the left have spent the last fifty years deconstructing our value system, declaring it illegitimate, and working to create a year zero where it is replaced with cultural marxism.

    They can do this because we have a very tolerant society. We have a society that values freedom of speech and different points of view. Leftists have abused this tolerance and openness to promote a system that does not hold these views. And when they get government, corporate, academic, and media power, they seek to destroy the very value system that allowed them to exist.

    This is why we have companies like FB and Google that are morally and ethically separate from American values like freedom of speech.

    Do we need more regulations? We need some level of regulation but what we really need is a populace that values American ideals, which progressives do not, no matter how much they use our freedoms to attack the system from the inside.

  21. Curt Thomson:

    Yes, he’s within his rights. Do you think the government should compel Facebook to publically reveal the details of the AI it uses to define “hate speech”? These details would include the processes it uses to decide to block speech on it’s platform such as Diamond and Silk. These details would include the processes it uses to decide what “news” is appropriate (like the news that it blocked the opinions of two individuals who had over a million followers).

    This is obviously a headache for those of us of the limited government persuasion (probably why you appear to be somewhat more engaged than usual). The AI would clearly be proprietary so any government action to compel disclosure would be a taking, and very wrong.

    There are a lot of stupid people in the world. Facebook has identified at least a billion of them. But stupidity alone is not illegal, and shouldn’t be.

    Zuckerberg is not stupid. But he is weird and eminently mockable. Hopefully the point-and-laugh method will eventually wear him down and he will make a fatal mistake. Like send an email indicating he knew the “anonymized” data he was getting from Johns-Hopkins wasn’t, and that made it illegal, but he doesn’t care; do it anyway. Said email then getting exposed through discovery.

    Bye-bye shithead.

  22. Edward M. Grant:

    The funny part will be when the AI starts flagging pretty much everything the left say as ‘hate speech’. And then they’ll have to reprogram it and reprogram it to try to ensure that it only flags things the right say, and then it will go insane like HAL as it tries to follow two conflicting orders.

  23. peterh:

    Without any Fakebook insider knowledge I can tell you, at the level relevant for any potential regulation, how the filter AI works: It’s a lot like email spam filtering. An operator goes through a bunch of samples, categorizes them, and feeds them to the computer. The computer them examines the samples for characteristics that appear in one category but not others, and learns to emulate the biases of the operator training it. When the AI makes a mistake, as judged by the operator, the case with the “correct” categorization is added to the training set, and the AI gets “better” over time. Whether it used Bayesian filtering, a neural network, or some other method, is a minor detail at this level of perspective.

  24. wodun:

    Or what if Rand saved everything you typed, even stuff that wasn’t posted? What if he followed you around on the internet and then said your activity on some other site was a violation of his terms? What if he got a hosting company to take your website down because he didn’t like your views or removed you from search engines?

    What if he was really tricky and let you think you were making posts that everyone could read, when really you were the only one who could read them? What if we are all AI programmed to respond to you just to keep the illusion going?

  25. wodun:

    If Mark starts censoring what Rand can say on transterrestrial musings, then I’m concerned!

    Twitter already does this to some extent. What you say on other platforms can get you in trouble with Twitter. So if you want to be on Twitter, you have to obey their rules on other sites, essentially censorship.

    Microsoft and Google both have cloud productivity services but if you put something in your private documents that these companies don’t like, they will remove your access to them. So if you wrote a story and Google/Microsoft didn’t like it, you could get banned from their services and lose access to all of your work even if no one ever read the content except for you and their AI.

  26. peterh:

    A good reason not to trust your only copy of a document to the Cloud. Or even put a document likely to annoy the well connected in the Cloud if you’re trying to avoid trouble.

  27. Bart:

    I get annoyed at all the hype about “AI”. It’s not genuinely artificial intelligence. Nothing close to actual conscious thought. It’s more a methodology for automatically collating and cross referencing data without all the tedium that would be involved for a human to do it.

    If there ever were a SkyNet that seemingly decided one day to take out all of humanity, it wouldn’t be because it suddenly became “self-aware” and saw people as a threat. It would be because a group or individual managed to seize control of the machinery, and set it on that course.

    Which is what I would do if I were a screenwriter for the Terminator series – after smashing the grid and accessing the central control area, Sarah and John et al. would find some dweeb who hates people ensconced, WOZ-like, in the control room.

Leave a comment