The World Will Be Our Skinner Box
thefrailestthing.com
thefrailestthing.com
I think there is a battle between two competing AI philosophies. One philosophy thinks AI should be central: all devices need to talk to the overmind for learning. The other philosophy thinks AI should be local: the device should learn on its own, kind of like a biological agent. I think the 'biological agent' form of AI is interesting when it comes to privacy of information. Only you and the device have the data.
The future of AI should be interesting with these two philosophies in mind.
The big question is whether we can build systems that communicate, but don't create vast disparities of information between overminds and everybody else. At this point the handful of information slurping mega-corporations have a huge information advantage compared with the average person. If nothing changes, this is the end of the humanistic world order based on the primacy of the individual brain.
I suspect is usually a mixture. For example Google Translate runs locally if you download a language pair, yet the training is not local. Google Gboard Swipe runs locally: in airplane mode I just added a new word to the dictionary and swipe recognised it, also contact names. However Google sends swipe information to a central location to improve the service...
These scales by which progress toward the future is measured are a key ingredient of social engineering. One task that Free Software Foundation type organizations could take upon themselves is to define a road map for future progress of technology that recognizes freedom and human dignity.
Everyone who's ever sat in traffic. It's a miserable experience and it is indeed a fundamentally better thing to be able to use that time in a more productive or entertaining fashion. Few people care if the steering wheel is entirely absent, so long as it's not needed and it's not in the way.
> The "progress" defined here is complete control over all actions of individuals in society by an AI god.
Who is defining progress that way? Everyone I know who wants a self-driving car wants it to go where they tell it to.
I'm sure many car enthusiasts care whether the steering wheel is entirely absent. Some people enjoy driving, albeit usually in times/places without much traffic. If someone who found driving therapeutic instead spends their commute time staring at a social media feed on their phone because that's their default action when they aren't sure what to do, would that be fundamentally better for them?
> Who is defining progress that way? Everyone I know who wants a self-driving car wants it to go where they tell it to.
Of course everyone wants self-driving cars to go where they tell it, but that is different from having control over where the car goes. What if the AI that controls the self-driving car takes 99% of people to their destination, but rounds up a few political dissidents and whisks them away to an unknown location. Most everyone would get what they want, but none of the passengers are able to control where the AI takes them. Like Stallman says, "Either the users control the program or the program controls the users."
Perhaps I was ambiguous. You're reading my intent backwards. Few people specifically want the steering wheel to be gone. They generally just want to not care about it.
> If someone who found driving therapeutic instead spends their commute time staring at a social media feed on their phone because that's their default action when they aren't sure what to do, would that be fundamentally better for them?
The car enthusiasts I know still don't like sitting in traffic. It's not the fun sort of driving.
> Of course everyone wants self-driving cars to go where they tell it, but that is different from having control over where the car goes. What if the AI that controls the self-driving car takes 99% of people to their destination, but rounds up a few political dissidents and whisks them away to an unknown location. Most everyone would get what they want, but none of the passengers are able to control where the AI takes them. Like Stallman says, "Either the users control the program or the program controls the users."
I don't know what to say in response to this. This feels over the top absurd to me. If your government is this fundamentally evil and amoral, I don't think driving your own car is going to protect you. There are already lots of ways to get rid of political dissidents.
The core problem is not the car, or who is driving it, but the fact that people need to spend a not insignificant amount of time traveling to and from the location or their employment.
It seems to be defined as convenience by any means necessary.
Umm, no, the "progress" is machines that I don't have to micro-manage anymore. Imagine an oven in your house that you had to constantly adjust to keep it at the desired temperature. No, that would be stupid, we've solved that problem decades ago. Now imagine a car that you had to control by hand with the necessary micro adjustments needed to keep it between the lines. No, that would be...well, it will be stupid, and a solved problem, one of these days.
"Hey, oven, make your interior 425F so I can bake a pizza."
"Hey, car, drive me to Dusty Strings in Fremont. Oh, why am I talking to you? You see it on my calendar, and you've been warmed up and ready waiting for me."
I don't know what dystopian future you're imagining, but I don't see the AI gods whisking me off to places I don't want to go. I see machines that finally, FINALLY, do what I want without me having to speak baby talk to them.
Oven: “Sorry mike. I can only go to 400F unless you upgrade your ovenware plan to $2.99 a month. Or use an ovenware family brand pizza! For every 3 ovenware brand pizzas you cook, you get a chance to win a free pie! Oh by the way on your trip to Dusty Springs tomorrow stop by Chad’s Bakery for their famous pies.”
"Hey, oven, make your interior 425F so I can bake a pizza."
Assuming that you didn't leave a dish in there the night before.
And that your child didn't hide a favorite toy in there... or a sibling.
I know you're using hyperbole here, and many people will disagree with what I'm going to write next, but that's exactly what I think the long term goal of technology should be. Not just individuals but all of our systems as well.
I'm struggling with this myself so I'm just echoing.
I guess if we're throwing around the term God it kind of loses it's juice.
God in this sense would be "thing that can do things nobody understands." In that sense, yes. Artificial General Intelligence (or maybe even Narrow AI) which can do things that humans can't understand for our benefit would look like "God" to most people.
Nobody can point to a reasonable causal chain of action for previously described "Gods." In the situation I'm advocating for, we would be able to do that - though likely still to a lesser degree than full interpretability.
If the self-driver becomes better than the best human drivers, there's limited gains in real safety if a 15-year-old with his learner's permit/drunkard/suicidal/homicidal/distracted person can still grab the wheel and steer it into a wall manually.
We may find that there are some opportunities that only become feasible if we remove the human factor. I could see a road network with much higher speeds for advanced self-driving vehicles, especially if you rely on the vehicles being able to communicate directly with each other and signaling mechanisms to keep traffic more orderly than a typical freeway.
There may also be some interesting business angles in it-- I fully expect that the feature after "we removed the steering wheel" is "we put a lie-flat bed, workstation or minibar where it lived."
Who hasn't delete Steam from their computer for a few weeks, thrown out junk food or set up a good workspace in order to promote a behavior in ourselves we want to cultivate, to be more like the person we want to be?
The important thing is that in those situations our environment remained under our control and conscious, and we need to ensure our personal and digital environment remain under our control, and its goals are transparent - and I am not sure companies like Google do or will always agree.
You have a motivation toward self improvement, you make a change to your environment, monitor it, and see what happens. You want to increase your physical fitness, awareness, education, life position, finances, etc. and you make corresponding changes where you and yours are the primary beneficiaries.
Google has a motivation to deliver value to their shareholders. They want to deliver more, so they take advantage of their privileged position in our lives and change our environment for us without asking, monitor it, and see what happens. They have far more capability for analysis and, considering the scale, far more financial motivation to make tiny adjustments that we would pass over in our own optimizations.
Is it necessarily evil or bad for us? No, it doesn't have to be, but I have very little faith in Google to make any choices that even border on altruistic when it's far more profitable to exploit us. Let's remember that fundamentally they are an advertising company, and that drives most of the money in the business; advertising isn't just ads as we know them, but fundamentally the business of changing your mind, suggesting things, and influencing your behaviour to make you exchange some money for some service or good.
When you have a privileged position of influence, and you're willing to sell it to literally anyone who signs up, there is clearly a lot of potential for abuse.
I think there's a case to be made here that Google's motivations can mirror those of the users and that it's in Google's best long term interest for that to be the case.
It seems obvious to me that a company that can be aligned with their customer's desires will have the most longevity, which may mean forgoing temporary, short term revenue gains.
I'm not so naiive to think that it currently is the state of things, or that it's simple for this to be the case - quite the opposite. Rather, my point is that there is no law of economics that guarantees users and corporations must be in conflict or that corporations will always do better financially by exploiting users to the users' detriment (or that a competitor that does that will beat one that doesn't).
But isn't there?
I think the core point of Meditations on Moloch[0] was that there kind of is, that in the limit, competitive environments will always sacrifice every value other than the one over which they're competing, and that an "AI god" would probably be the only thing that's strong enough to break these chains.
--
[0] - http://slatestarcodex.com/2014/07/30/meditations-on-moloch/
However those don't conflict with my point that there is nothing preventing a firm and the sum of their users from having strictly aligned goals in the context of their relationship.
Competition simply means that there could be other firms who may be able to align their goals with users goals more tightly, and run the other firm out of business. Much more to be probed here, like consistency, short and long term goals and desire/goal uncertainty on the part of the user and firm. None of those seem to be intractable issues.
Unless your organization uses Google Apps, or standardizes company cell phones on Android, or uses Google Cloud for hosting things...
I want to see employees have more negotiating power, but you’re being ridiculous. Yes, there’s a whole segment of employees who are exploited by their employers because they are financially vulnerable and may not be able to easily get another job. I’ve worked some of those jobs. They’re mostly not the types of jobs where there’s a lot of concern because their boss makes them use Google Apps.
And going down this road, there’s a huge variety of things you can say your boss “forces” you to do, where it becomes meaningless.
My boss “forces” me to use this brand of office furniture. My boss “forces” me to be here during business hours. My boss “forces” me to do work, for crying out loud! The inhumanity!
You seem to think that if anyone has any ability to cause you inconvenience or even harm if you don’t do their bidding, they’re forced you.
>But the Panopticon was also a laboratory; it could be used as a machine to carry out experiments, to alter behaviour, to train or correct individuals. To experiment with medicines and monitor their effects. To try out different punishments on prisoners, according to their crimes and character, and to seek the most effective ones. To teach different techniques simultaneously to the workers, to decide which is the best. To try out pedagogical experiments – and in particular to take up once again the well-debated problem of secluded education, by using orphans ...
It's interesting to note that the argument made in this post is exactly what Deleuze said two decades after Foucault [1], that the mechanisms of "discipline" are moving outside of enclosed spaces because of the ever-longer trails of data that we leave everywhere.
I've already been calling Internet fads, echo chambers, and meme outbreaks "Borgsongs." I seem to recall this term from STTNG for the omnipresent hive signal experienced by Borg.
A short story even asked the question, could you engineer a meme/song so powerful that it could literally make you go insane? We know this stuff exists (brainwashing, cults, ...), but can you super concentrate it?
At some point we are going to be forced to start regulating persuasion.
We burned stuff long before modern steam turbines and internal combustion engines...
Now we have another qualitative step - never before people were actually connected 24/7, with constant Internet activations coming to our pocket.
Of course, the risks are different when you're trusting a device from a large corporation to take on a role that a human would do.
I wouldn't buy one, but some people will decide it's worth it, and I don't think they are losing their freedom by giving orders to a machine.
The device doing its job can have my information, but sending it back to the mothership, even if it's for the purposes of improving the service for such devices in general, then it's the equivalent of a real assistant babbling about my private stuff to their colleagues, which is immoral and a fireable offence.
Google at least makes this explicit with "Ok, Google".
'You've effectively hired KGB/CIA to work in your house, the human assistant is just the interface'.
Banks are more regulated than Internet companies, though, so maybe there is something to be done there.
You've hired a tiny, tiny string of server activations, through which your voice gets interpreted and turned into query to be run, with some amount of side effects that may affect your device directly. Just because some people cannot see through the abstraction of a company name, doesn't mean we should let it be the accepted view.
"Sure, I am too aware of these techniques and mentally strong to ever be influenced by marketing, but what about the rest of humanity? Google will control their minds like a virus and soon destroy the world!"
I like to think (and
the sooner the better!)
of a cybernetic meadow
where mammals and computers
live together in mutually
programming harmony
like pure water
touching clear sky.
I like to think
(right now, please!)
of a cybernetic forest
filled with pines and electronics
where deer stroll peacefully
past computers
as if they were flowers
with spinning blossoms.
I like to think
(it has to be!)
of a cybernetic ecology
where we are free of our labors
and joined back to nature,
returned to our mammal
brothers and sisters,
and all watched over
by machines of loving grace.
by Richard Brautigan (1967)
Do you want to pay for all the hundred sites you visit in a year?
The problem privacy advocates have is forging a link between the abstract and the concrete. I don't really care that Experian compiles a list of my credit data. I really do care when they fuck up and leak it publicly. The tricky thing is that the former is a prerequisite for the latter, so I should care about that, too.
Exactly. That's why I put "privacy" in quotes. I think people are smarter than you give them credit for. I don't give out the password to my email account, but do I care that much if Google has some machine learning algorithm with a bunch of data points buried in there that I use the word "NYC" a lot? No, I really don't care. And yeah, they might get hacked and leak my emails to the world. That would suck. But that risk is worth it to me, and judging by the hundreds of millions of people with Gmail accounts, I'm not alone.
I generally think open dialogue is a good thing, but your comment just bolsters those in positions of power to the detriment of everyone else for literally nothing other than reducing your own minor personal annoyance. Want better ads? Mail them a dossier on your life story and leave the rest of us out of it.
Would you prefer to hide:
-recordings of your sexual encounters from your family?
-the raunchiest porn you've ever looked at from your co-workers?
-your live GPS location from home burglars?
-your bank account balance from friends?
-your tax filings from scam artists?
-your off-color remarks you made 15 years ago while drunk from the person conducting your job interview?
-the drugs you tried in college from law enforcement?
-The minor legal transgression you committed from the person looking to shut down your political protest?
-the psychological trigger even you were unaware of from the deep learning algorithm designed to addict you to gambling.
-your sexual orientation from the autocratic government of the vacation spot you visited?
The only people who truly have nothing to hide are those who truly have nothing to lose. If that's you, then I'm sincerely sorry.