5 ms·

AI could be “a fascist's dream”

18 points by malloryerik 10y ago

eli_gottlieb 10y ago
Oh for fuck's sake. AI could also be a liberal's dream, or a socialist's dream, et cetera. Stop trying to treat technology as having some innate political orientation, and take responsibility for the degradation by humans of our politics.
beaconstudios 10y ago
Yet another article equating nationalism and conservatism with white supremacy and fascism? Pathetic.
digitalsushi 10y ago
I'm not in the field of machine learning, or AI or even know how to use these terms with any expertise. It has, however become a lingering thought that there will come a day where everything we have clicked, liked, linked, saved in email, posted on walls, left in caches, and perhaps even said in audio or mouthed in video, will eventually be distilled and converted into some hashcode that instantly exposes our true selves. Our private biases, our religious, sexual preferences, our kinks, our racisms, all our private inner thoughts are possibly one powerful calculation away from being public. And I dont know if an open person can appreciate the type of hell that would be for a shy person.
- jasonkostempski 10y ago
  For the existing generation it would be an obvious issue, but I wonder if shy people would even form in such an environment.
  - mercer 10y ago
    Mind elaborating on that? It strikes me as an interesting thought.
padobson 10y ago
Is there some mad outbreak of fascism I don't know about? Anyone who regularly reads the Guardian and takes it seriously is probably locked up in a seed bunker somewhere. I certainly agree there's been a slow erosion of civil liberties in the US since 9/11, but I've yet to see an epidemic of doors kicked in for outspoken opinions against the state. Maybe calm down on the fascism talk.
mtgx 10y ago
Another relevant story from today: https://www.wired.com/2017/04/courts-using-ai-sentence-criminals-must-stop-now/ https://www.wired.com/2017/04/courts-using-ai-sentence-crimi...
clort 10y ago
I don't know about AI but the deep learning algorithms the article mentions can certainly be used for evil purposes. In truth, the people being governed by algorithms need to learn that computer says does not imply correctness and that the power is in the people governing whatever the tools they use. Actually, I think that genuine AI which is smarter than the current set of politicians could be a benefit since it should not have the same limitations and actually be able to think things through to a generally better outcome, rather than pandering to its own biases and prejudices (see: The Culture)
- vorotato 10y ago
  AI trained from human's reading material are prone to the same biases that humans are. Just a forewarning to anyone going down this path.
denzil_correa 10y ago
We do really need accountability and explainability for ANY algorithm - AI or not. I see a lot of organizations wash their hand off the results by pointing fingers : "Hey, we didn't do it - the machine did it". It's a potential disaster in the making - not just for people but also for technology itself. The potential impacts of an algorithm should be studied and a decent attempt has to be made towards this aspect before you permeate them in every minute of your lives. It's a disaster waiting to happen and once it does, people will mistrust technology.
- dragonwriter 10y ago
  There's an old saying (well, old as such things go in computing, at least) that seems relevant here: “every error blamed on a computer involves at least two human errors, including the error of blaming the computer.” But that seems to be the point of the article; it's not blaming the technology, just noting that the technology can be misused that there areexisting social trends that are visible at the same time the technology is becoming more widespread that many people would have reason to be concerned about leveraging the technology. AI has no inherent ideology, but it makes it possible for the user's ideology to be applied more effectively and with less opportunity for dissenters whose cooperation would otherwise be necessary for implementation to undermine it.
malloryerik 10y ago
May I just ask, on what grounds was this flagged?
- Mz 10y ago
  Without looking at the article, I will suggest one possibility is that people are reacting to the inflammatory, click bait title. I have had perfectly good articles flagged to death due to poor titling. In one case, I deleted it, asked for a title suggestion from the mods and resubmitted it with their permission and things went significantly differently the second time around. The HN crowd is pretty sensitive to bad titles.
  - malloryerik 10y ago
    Right, I think I hadn't imagined how sensitive some, apparently on the political right, would be to the word "fascist." The article is about a Principal Researcher at Microsoft's SXSW presentation, "Dark Days: AI and the Rise of Fascism." The title on HN was merely a condensed version of the article in The Guardian. (I was going to link to the Financial Times' article but there is a paywall. The Guardian is a top global newspaper though, and was being attacked here as a mere hippy rag, whose readers must reside in "seed bunkers," so, wacky believers in climate change, much the way the New York Times is attacked as fake news.) Seems to me that those who were offended had their amygdalas tripped by the word "fascist" because these days when they see that word, they feel as if it's being used as an attack against them, and the larger conversation is thereby hijacked. Meanwhile a highly-related article about AI being leveraged in China to adjust credit scores according the way that people charge their phone batteries wasn't flagged. So it seems that you're right.

18 points by malloryerik 10y ago

Oh for fuck's sake. AI could also be a liberal's dream, or a socialist's dream, et cetera. Stop trying to treat technology as having some innate political orientation, and take responsibility for the degradation by humans of our politics.

beaconstudios 10y ago

Yet another article equating nationalism and conservatism with white supremacy and fascism? Pathetic.

digitalsushi 10y ago

I'm not in the field of machine learning, or AI or even know how to use these terms with any expertise. It has, however become a lingering thought that there will come a day where everything we have clicked, liked, linked, saved in email, posted on walls, left in caches, and perhaps even said in audio or mouthed in video, will eventually be distilled and converted into some hashcode that instantly exposes our true selves. Our private biases, our religious, sexual preferences, our kinks, our racisms, all our private inner thoughts are possibly one powerful calculation away from being public. And I dont know if an open person can appreciate the type of hell that would be for a shy person.

jasonkostempski 10y ago
For the existing generation it would be an obvious issue, but I wonder if shy people would even form in such an environment.
- mercer 10y ago
  Mind elaborating on that? It strikes me as an interesting thought.

padobson 10y ago

Is there some mad outbreak of fascism I don't know about? Anyone who regularly reads the Guardian and takes it seriously is probably locked up in a seed bunker somewhere. I certainly agree there's been a slow erosion of civil liberties in the US since 9/11, but I've yet to see an epidemic of doors kicked in for outspoken opinions against the state. Maybe calm down on the fascism talk.

mtgx 10y ago

Another relevant story from today: https://www.wired.com/2017/04/courts-using-ai-sentence-criminals-must-stop-now/ https://www.wired.com/2017/04/courts-using-ai-sentence-crimi...

clort 10y ago

I don't know about AI but the deep learning algorithms the article mentions can certainly be used for evil purposes. In truth, the people being governed by algorithms need to learn that computer says does not imply correctness and that the power is in the people governing whatever the tools they use. Actually, I think that genuine AI which is smarter than the current set of politicians could be a benefit since it should not have the same limitations and actually be able to think things through to a generally better outcome, rather than pandering to its own biases and prejudices (see: The Culture)

vorotato 10y ago
AI trained from human's reading material are prone to the same biases that humans are. Just a forewarning to anyone going down this path.

denzil_correa 10y ago

We do really need accountability and explainability for ANY algorithm - AI or not. I see a lot of organizations wash their hand off the results by pointing fingers : "Hey, we didn't do it - the machine did it". It's a potential disaster in the making - not just for people but also for technology itself. The potential impacts of an algorithm should be studied and a decent attempt has to be made towards this aspect before you permeate them in every minute of your lives. It's a disaster waiting to happen and once it does, people will mistrust technology.

dragonwriter 10y ago
There's an old saying (well, old as such things go in computing, at least) that seems relevant here: “every error blamed on a computer involves at least two human errors, including the error of blaming the computer.” But that seems to be the point of the article; it's not blaming the technology, just noting that the technology can be misused that there areexisting social trends that are visible at the same time the technology is becoming more widespread that many people would have reason to be concerned about leveraging the technology. AI has no inherent ideology, but it makes it possible for the user's ideology to be applied more effectively and with less opportunity for dissenters whose cooperation would otherwise be necessary for implementation to undermine it.

malloryerik 10y ago

May I just ask, on what grounds was this flagged?

Mz 10y ago
Without looking at the article, I will suggest one possibility is that people are reacting to the inflammatory, click bait title. I have had perfectly good articles flagged to death due to poor titling. In one case, I deleted it, asked for a title suggestion from the mods and resubmitted it with their permission and things went significantly differently the second time around. The HN crowd is pretty sensitive to bad titles.
- malloryerik 10y ago
  Right, I think I hadn't imagined how sensitive some, apparently on the political right, would be to the word "fascist." The article is about a Principal Researcher at Microsoft's SXSW presentation, "Dark Days: AI and the Rise of Fascism." The title on HN was merely a condensed version of the article in The Guardian. (I was going to link to the Financial Times' article but there is a paywall. The Guardian is a top global newspaper though, and was being attacked here as a mere hippy rag, whose readers must reside in "seed bunkers," so, wacky believers in climate change, much the way the New York Times is attacked as fake news.) Seems to me that those who were offended had their amygdalas tripped by the word "fascist" because these days when they see that word, they feel as if it's being used as an attack against them, and the larger conversation is thereby hijacked. Meanwhile a highly-related article about AI being leveraged in China to adjust credit scores according the way that people charge their phone batteries wasn't flagged. So it seems that you're right.