Article: Can AI Help Gender Diversity Help AI?

http://motherboard.vice.com/en_uk/read/can-ai-help-gender-diversity-help-ai

Cognitive and Psychological Effects - Data, Numbers, Machines

That Facebook numbers trick is so obvious right after someone tells you about it. I never gave much thought to the reasoning behind that, which in hindsight seems really odd. That whole racking up likes thing does kind of give you a warm fuzzy feeling, but the whole 1,682 friends on Facebook thing is a little nuts. I've never even met that many people, let alone that many people I would call, even in the loosest of definitions, friends.

Of course, one of the results of having taken more than my share of social psychology-related classes is that I'm familiar with the concept of Dunbar's number,  If you aren't familiar, it's a social science concept that suggests we have a cognitive limit on the number of stable relationships we can manage at one time, and that number is 150. So basically, however many friends you have, you have 150 friends.

Anyway, I installed the Facebook Demetricator because I'm curious. Does it actually make Facebook a more pleasant and relaxing experience? We'll see.

I have a pretty fair comprehension of algorithms and was still a little shocked at just how far they go. I mean, seriously. I'm a Mac user and there's no way I'm going to choose anything that can be classified as more expensive because I'm also a broke college student.

The implications of their capabilities, though. I always thought of algorithms as basically helpful if maybe slightly irritating sometimes. Helpful in the sense that they're shortcuts, irritating in that sometimes they don't work exactly right and Netflix not only can't remember what episode of Gilmour Girls you were watching last night but it keeps trying to show you cheese documentaries.

It stands to reason though that if machines can learn, they're going to pick up things they shouldn't or process normal things in ways they shouldn't. Kind of like toddlers. Maybe exactly like toddlers, mimics who don't really comprehend that they're saying something they shouldn't say, they've just picked it up from the world around them. Or been taught to say it by someone who thinks they're funny but is really just a little bit evil.

Teaching them seems like a really large job, but regulating the exploitation of them seems like an even bigger one. I found this whole concept kind of fascinating, in a  horrifying sort of way, and my horror was not dispelled when I ran across an article about the Chicago PD's heat list.

Basically, they got this really big grant ($2 million) to set up and experiment with an incredibly high-tech piece of computer gadgetry that has allowed them use algorithms to profile people based on a set of criteria that included geographic area, volume of emergency calls placed, previous interactions with the PD, and a whole slew of other items that generated them an index they called the heat list.

Not surprisingly, there were people on it who never committed a crime or had any kind of run in with the police whatsoever.

Seems legit, though.

Psychological and Cognitive Effects

I already know I’m really bad at multitasking. I'm pretty sure I got to the end of the test but then again, I think they were supposed to give me results.  

I have a husband, two kids, and three dogs. Two birds. Couple turtles. A house that refuses to clean itself, that kind of thing. I figure someone with a really loose definition of multitasking might say that I practice it, but they’d be wrong. My personal method would be defined less as multitasking and more as starting a hundred and six different tasks, one right after the other, and then migrating from task to task. For days on end. It’s an aimless sort of process and doesn't save me any time, but it does create the illusion that I’m getting more done than I actually am. 

If that New York Times article on faking it is any indication, that 2% statistic is not a misprint. For the most part, we humans like to save time and cognitive resources by relying on our automatic responses. Hey, it serves us well most of the time. 

I’m no supertasker by any stretch of the imagination, but I probably do fit Hemp’s description of the information addict. Addict is such a strong word, though. I really prefer the more informal "information junkie". 

In my experience, most EMAC people are information junkies. We’re the Internet curators and fact checkers and social media managers. Information collectors and dispersers. 

In a class I took a couple years ago, some fellow students conducted a sort of mini-study on the sharing of misinformation. They surveyed students in the ATEC, JSOM, and SU buildings and found that most respondents had at some point shared information that was later proven incorrect. Their responses went something like, "We never imagined it wouldn't be true, and anyway, who cares?" 

EMAC students were the exception. To us, that sort of gaffe is practically criminal. Our battle cry is, "Hang on, I'll look that up!" We know the information is there for the taking. We can't even help ourselves. 

If we’re experiencing information overload, the situation must be truly dire. 

We’re all full-fledged participants in what social media scholar danah boyd calls the ‘always-on lifestyle’. When asked how many hours a day she spends online, boyd can’t come up with a number because she reaches a stumbling block immediately: What exactly counts as online? 

Count up the hours you sleep, she suggests, and add them in with the time you spend in the shower. Hmm, do you scroll through Facebook when you can’t sleep? Tweet from the dinner table? Would those be ‘on’ times, then? 

Our environment is complex. It’s twisty and elaborate and hurls us piles of data nonstop. Adding more on top of that was bound to throw a kink or two our way. 

Being connected to the network constantly has changed the way we do a lot of things. According to the research we’ve seen, print news isn't stumbling along on its last legs yet, but the Internet has still markedly changed the way we consume news. 

Pentina and Tarafdar's study on social media’s role in news consumption found evidence that information overload doesn’t necessarily dictate that people actually consume it. 

As the Times article has already pointed out, we decipher bits and pieces and pick up summaries and memes and others’ remarks. These bits and pieces can provide us with enough about a given subject that once we've internalized them, we start to wonder where the wisdom is in gathering knowledge for ourselves. 

We're basically a society of skimmers. At the very least, the gist of anything is all we really need to get by. 

Privacy in the Digital Age

The article in the Guardian, "The death of privacy" , mentions the Facebook contagion experiment. I'd forgotten all about that. That quote in the article from that guy at MIT, Hidalgo, about how the Facebook News Feed is just like sausage, that really made me giggle. It's so true but we don't really let ourselves think of it that way if we can help it. Facebook took a pretty good beatdown in the media and from the public over that one. They apologized, of course. And then they amended their terms and conditions to allow data to be used as research. After they had already used it as research. 

Now that I didn't know. Sausage. Hmm. 

 And then I remembered, hey, didn't OkCupid do that same thing? Indeed they did. And then blogged about it.  Christian Rudder, data scientist and co-founder of the site, explained that well, OkCupid doesn't exactly know what it's doing. No website really does, he says. So curiosity got the better of them, I suppose. Those three experiments Rudder blogged about actually netted them some interesting data. One, that most of a user's rating is based on the photo they use on their profile, not the accompanying text. 44% of the time. Two, people will blindly trust the site to tell them who is or isn't a good match. 

According to that Pew Research study on what Americans think about data collection and security, we don't really care for this kind of thing. And apparently, the better we understand it, the less we care for it. Obviously there are excellent reasons for this. Don't we own our own data? We should have a say as far as what we share or don't share, right? Right. But I think it's sort of like that stages of grief thing. 

We’re appropriately disturbed. At first. Then we flood social media with sarcastic memes. Then we roll our eyes and go on with life because, really, what can we do about it anyway? 

A spokeswoman for Facebook was quoted as saying that “when someone signs up for Facebook, we’ve always asked permission to used their information to provide and enhance the services we offer…” And I’ll stop there. Let it sink in. Enhance the services they offer. 

We are always sold that line. It's just a little pinch. It's for your own good. 

We know better, though. The Annenberg study discussed in "Sharing Data, but Not Happily" certainly points to that. There seems to be a failure to communicate clearly there. Companies think that the freebies and discounts they hand out to consumers as consolation prizes are payment enough for the data they're harvesting, but the consumer, for the most part, doesn't agree. Not altogether surprising. We don't want to pay for free wifi with our personal information. 

We don't care to be tracked. We don't want our data gathered and sold to the highest bidder(s). We REALLY don't want anything to do with online advertisers. They're shady. 

But then again, it seems that most every online entity is. Social media. Online advertisements. Dating sites. News organizations. Especially them. They're operating under the misguided assumption that we appreciate their enhancement of our experience, and that the offer code we got for free shipping at Name Your Favorite Online Retailer was payment enough. 

Relationships in the Digital Age

     Most of the time (okay, all of the time), my Facebook/Twitter/Instagram feeds make me incredibly grateful that I am 36 years old. I navigated my way through middle school and high school without social media. Without a smartphone. Or Google. Well okay, I do feel as though my high school education could have stood a little Google in it. But other than the lack of Google and resultant necessity of gathering research from actual books, I think we managed to do okay. It probably doesn’t hurt either that my husband and I have been together for 17 years so our relationship definitely pre-dates any Liking, Following, or Commenting angst. 

     Millennial teens are the pioneers of digital relationships, both platonic and romantic. If the online activities of my teenage son and his friends are any indicator, it’s an experience that’s both exhilarating and exhausting. They’re always on. 24/7, in fact. Sometimes wrestling the electronic device from his hand seems inevitable, it’s like a limb or something. They do socialize in person, seeing movies, hitting the swimming pool in the summer, attending those always excruciating school-sponsored social activities. They’re getting together, in the same room, ostensibly interacting with one another, but they’re doing it with their iPhones glued to their hands. And the kicker is this: I never let my son leave home without it. How else will I know when he needs a ride home? Find my iPhone isn’t a very good checker-upper if the iPhone in question is sitting on the kitchen table at home. 

     In my day, we didn’t walk to school and back both ways uphill in fourteen feet of snow, our backs pelted with soccer ball sized hail. No. In my day we made due with phone booths. And they cost $.25 a call, too. 

     The phone booth offered us another advantage, one I’m only just beginning to see the value in. We just picked up the phone and spoke. The same words we always used. A few months ago my son needed a ride home from band practice. I gave him the usual spiel about where to be, don’t forget your backpack, etc. He replied with ‘Akai’. It took me a minute to decipher that. Okay. He means okay. He could have just said ‘okay’. Or even ‘ok’. 

 

I caught the gist of that study on the role of the period in computer-mediated communications in a few different online articles, and I have to admit, it kind of threw me. I use punctuation. In texts. Facebook posts, emails, to-do lists, they all have periods and commas where periods and commas should go. 

HOW? This made no sense to me. That’s sloppy, right? It’s like using ‘UR' instead of ‘your’ or ‘ppl’ instead of ‘people’. Smartphones relieved us of the 140-character minimum, and, Twitter aside, web-based social networking doesn’t limit how much you can say, so why not spell? Punctuate? Is the period just too final? Did threaded messaging present us with both longer message capability and a fresh new way to feel socially awkward? Maybe the sentence-final period is the Millennial version of “You hang up.” “No, you.” “You first.” “Same time, count of three!”  

 

So I started paying a little more attention to my son and his friends’ interactions online. As a rule, they don’t punctuate much. They misspell things on purpose. They use emojis to form entire sentences. Frankly, it was kind of fascinating. They’re interacting, they’re socializing. They’re creating fan fiction on Wattpad and microblogging about anime and Minecraft on Tumblr. 

These are the new digital social cues. I felt like I needed some kind of ‘For Dummies’ handbook at first, but watch long enough and you figure out that not only are they getting everything from stories to daily chit chat across to each other, they’re imparting joy, anger, sadness, and sarcasm to each other.  In a lot of cases, it's a pretty decent illustration of the concepts Julie Beck discusses in her article on depression and happiness spreading like contagion through the nodes of social networks. Sure, it can go both ways. But that phenomenon is clearly visible in the online realms of Millennial teens. 

Economics of Digital Media

Privacy is a more valuable commodity in the digital age. We’ve pretty much resigned ourselves to the notion that our government views our personal privacy as a privilege rather than a right and that we’ve been standing in the proverbial corner for quite a while now. 

There’s a pretty pervasive “we give up” attitude regarding government data collection. We care, of course, and we don’t like it, but it isn’t within our power to resist it. This overlying sense of Big Brother watching us ironically doesn’t seem to make us any more cautious. But it’s that resigned attitude that takes the sting out of the implications of things like store discount  card programs connected to our email accounts and cookies that follow us around, noting the things we like to click on and springing them on us as we search Google or scroll through our Facebook newsfeed.

The ever-evolving sophistication of online advertising techniques paint a pretty clear picture. Our online behavior and browsing habits have come to be viewed as a sort of abstract product. Advertisers, marketers, and business owners want to get at that information because it represents profit. And if more than one agenda can be addressed in one fell swoop, that’s even better. Our privacy is essentially the currency of the digital age.

Exploitation of our privacy in the name of profit increases has, inevitably, drawn attention to the ethical aspect of practices like native advertising. We’ve all experienced it at one point or another, when we search for a product or service via Google or on Amazon or eBay. Ten minutes later, upon logging into our Facebook accounts, the ads we see feature the same sort of products we’ve been perusing elsewhere. It’s sort of novel at first, and occasionally I’ve got to admit it’s been helpful. But it doesn’t take long for the creep factor to set in. 

Take for example the father of a teenage girl who, irate that the store had sent his daughter baby and pregnancy tailored advertisements in the mail, demanded an explanation. The explanation was a simple one. Target’s consumer tracking tools had simply honed into the fact that the man’s daughter was indeed pregnant.

And it gets better. 

According to a November article in the Atlantic, our smart technologies aren't just communicating with us, they're talking amongst themselves. And as they talk, they're providing valuable insight to advertisers about the users of those devices and the preferences they exhibit. 

Increasingly, websites are providing a disclaimer notifying visitors that the site is employing the use of cookies and this is often accompanied by an opt-out option. The interesting thing about this is that the wording almost makes it seem as though agreeing to be tracked is in the viewer’s best interest. “This website uses cookies to improve your experience” is far more persuasive than “Warning! We allow third parties to track you and your data may possibly be sold to the highest bidder!” 

The advent of these opt-out options on our US websites likely stems from the implementation of of the EU’s “cookie law”, which was intended to provide consumers with more options for data privacy. 

It shouldn’t have been surprising to me that news sites represent the most prolific use of third-party trackers, but it did, for exactly the reasons outlined in the Libert/Pickard article.

We expect a higher standard of ethics from our news providers. This makes sense, because as we’ve established, people seek out their news from certain sites for specific reasons, and one of the most compelling is that we’ve decided we trust the information we’re receiving there. 

Makes them seem a little greedy, though, doesn’t it? According to the Pew Research industry breakdown, news media represents billions of dollars in revenue. Maybe there’s a paradox there. Is the news industry’s impressive profit a direct and clear result of its use of third-party trackers? Research shows, for example, that the print segment of the news industry isn’t as close to its deathbed as we’ve been led to believe. 

Reading Response - Digital Media Users: Who is Using and Paying for What?

I don’t really consider myself to be a Millennial. Caught in that awkward stage between Generation X and Millennial, we have sort of experienced the best (and worst) of both worlds. As a late-90s high school graduate, my typing education took place on an actual typewriter. When we had to research anything, we went to the library. The actual building, not the nifty little website that I know and love today. And as an avid reader, I owned piles and piles and stacks of books in print. Honestly, I had no plans to change my ways. Except, obviously, tossing that behemoth typewriter in favor of a shiny new Windows PC. I loved the library. I preferred to read in print because I read, and reread, and mark and mutilate my books, and the scribblings inside may or may not have something to do with the book in question. 

When I returned to school for my bachelor’s degree in 2013, obviously things were different. While I was out, Google Scholar and online libraries had become a thing. And I embraced them wholly. I quickly learned that in the EMAC program, professors tended to assign digital readings, both because of the ever-changing material and because, well, they’re just cool like that. Print textbooks are expensive, after all. 

I had trouble with it, so Wolf’s theories on digital’s effect on the brain ring with familiarity for me. At first, I had to print a hard copy of everything. To be clear: I had trouble with digital reading, it seemed as though the process (or maybe the ritual) of reading in print was indelibly printed on my brain. As I discussed above, I don’t really consider myself to be a digital native. My preference for print was simply a throwback. But, after three years of reading largely digital copies of everything from academic papers to pop culture articles to novels, I find myself having to shift a gear or two in order to read and comprehend digital vs. print. 

As for Baron’s examination of student preference between print and digital, I tend to think that students’ preference of print because it’s “easier to follow stories” may simply come down to the level of distraction afoot on the average student’s laptop or tablet. Having a reading assignment or novel open and in progress on a device is no guarantee that it won’t be accompanied by twelve other tabs, each one containing something different and probably equally fascinating. 

Clay Shirky, NYU professor and Internet guru, actually has a few things to say about digital distractions. In an article originally published in Medium, Shirky discusses why, in spite of the fact that he is “an advocate and activist for the free culture movement” and therefore “a pretty unlikely candidate for internet censor”, he began prohibiting the use of laptops, tablets, and phones in his class. His reasoning for this, in spite of the fact that he viewed the keeping of his students’ attention as a challenge, was the simple fact that multi-tasking makes for subpar quality cognitive work. 

I can’t speak for anyone else, but when I’m attempting to read something on my laptop, tablet, or phone, multi-tasking is exactly what I’m doing, almost without fail. And this, just as much as scanning and the advent of nonlinear reading, is responsible for difficulties relating to staying on track with comprehension. And if, as Wolf suggests, we are in need of a bi-literate brain, the need for better comprehension of digital material should go hand-in-hand with better comprehension of print material.  

Being a 'print native', I guess you'd maybe call us, the Ramen Noodles Theory made complete sense to me. Of course I read many, many things online. Magazine articles, local news, tidbits from Refinery29, blogs, homework assignments, it's a lengthy list. But I'm old-school too. I pay for print magazine subscriptions, preferring them to the digital versions. I still get the local newspaper delivered every day. The contradictory nature of the Ramen Noodles theory stacked against the seemingly inarguable statistics presented by Kaiser in The Bad News About the News gave me pause, though. I had to stop and consider the distinct probability that they are resting side by side. Online news is prolific, and it doesn't have to come from the traditional sources.  

Reading Response Week Five - Information Seeking, Polarization, and Digital Democracy

About a year ago I took a topics in EMAC course called Social Networks. Most of the coursework revolved around social networks in the more organic sense, with social media and digital networking not quite an afterthought but more like a close second. That makes sense, though, I guess, since to understand the nuts and bolts of how relationships operate online you first should understand how they operate IRL. 

Anyway, one of our assignments for that class was to watch a TED talk given by Eli Pariser in which he warns us to beware of ‘filter bubbles’. Admittedly this is a concept that I probably should have been aware of before 2014, but I wasn’t, and I was equally horrified and amazed by the implications of it. 

Basically what Pariser is warning us about is the algorithms that so conveniently give us just the things we want to see online. One of his comments, about how your Google search results vary depending on which account/computer you’re using sent me searching on every computer in the house to test his theory. 

Huh. He was right. A particular Google search on my computer didn’t match exactly with the search on my son’s computer, or my husband’s. Not dramatically different, but still, different. Our computers, our Facebook accounts, and soon Twitter too will all be telling us exactly what we want to hear. 

On the surface this seems like it would be a great thing. Block out all the noise and only see content that’s relevant to you. So why beware? 

Those articles on political polarization are a pretty good answer to that question. The Pew study tells us that as a whole, politically, the American public is moving further and further from middle ground. Rights go further right, lefts go further left. This isn’t good news for politics because distance makes it harder to meet in the middle. Without going too much into my own political beliefs, I will say that this information made me stop and look at myself, particularly that part about how on both sides people believe that the policies of the other party are “so misguided that they threaten the nation’s well-being”. 

Do I believe this? Yes. Yes, I do. Why? Because of the information I’m getting online? Probably. That’s where I get my news. And while I don’t get all of my news on Facebook, I do often start there. Facebook, where it seems I’m surrounded by likeminded individuals who believe the same things I believe and who are interested in many of the same things I’m interested in. I suppose that makes a compelling case for the theory that Facebook use is polarizing. 

It’s a problem because it does us good to see opposing views. It’s good to know how the other side thinks. You never know, there may be that odd thing in there somewhere you actually agree with. Sure, it can be infuriating, but a little passion is healthy, right? 

After all, if the sentiments you come across defy all logic and sincerely make you question the ability of the sentiment's possessor to successfully tie their own shoelaces, just remember that there is an actual condition in which the dumb are simply too dumb to know they are dumb, and it's possible that person is suffering from it. 

Reading Response Week Four

A few years ago, Donald Trump Tweeted “I love Twitter…it’s like owning your own newspaper—without the losses.” Since Howard Dean in 2004, politicians have been learning to harness the power of social media. Twitter is an increasingly viable tool for circumventing the mainstream media. Rather than depending on a news organization to share the information they think is important, politicians can now use Twitter to create a new kind of sound bite - the Twitter bite.

To know that if you make enough waves or share information that is compelling enough on Twitter, you can catch the attention of mainstream news outlets, you only have to look to Trump’s digital campaign strategy. 

That journalists can harness the power of Twitter to engage audiences, share more information with a larger audience and simply don’t do so strikes me as being a sort of uncompromising neglectfulness. 

Yes, it’s true that people are always going to need news, but if they aren’t satisfied with what they’re getting from the more mainstream journalistic sources, they can go elsewhere with ease. 

Case in point: Paul Horner. His particular brand of ‘news’ is terrifying. It’s terrifying because there are people out there in the world who take him and others of his ilk at face value. He’s making his success on telling people what they want to hear. Of course, the problem with that is, what they want to hear often coincides more with fairy tale than truth. 

“I don’t know, Google it” is a phrase I utter to my children at least two or three times a day. When I was growing up, the Internet as a household staple was still fairly far up ahead in the distance. If we wanted information, we took ourselves to the library. Another refrain that often follows “Google it” is “There isn’t any reason not to know, the information is there for you to find.” 

Checking your facts is just the smart thing to do. 

In the third quarter of 2015, Twitter had a little over 300 million monthly active users. Facebook, by comparison, boasts over a billion monthly active users. by themselves, these numbers seem pretty self-explanatory. But, factor in the embedded Tweets seen elsewhere on the web, and Twitter sees over a billion unique viewers every month. 

This upward trajectory is what mainstream journalists should note, and take advantage of. Twitter utilization in the upcoming presidential election has given the medium a jumping off point to assert itself as the go-to site for breaking news coverage and real-time interaction with readers and watchers. 

Ethical Decision-Making - Week Three

Rosenberg’s discussion on the resolution of moral disputes kind of made me yawn at first. I didn’t necessarily disagree; on the contrary, what he was saying seemed more like common-sense type stuff. But his points on politics and religion caught my attention and made me dig a little deeper. Rosenberg says that political disputes occasionally have a moral aspect but mostly the arguers (Republican, Democrat, etc.), share the same ends but don’t agree on the means. But, in a foggy sort of way, he’s also saying that political disputes don’t always have a religious aspect, sometimes they do, and in that case, the political dispute becomes a moral dispute. Am I reading that right? 

Anyway. I like politics. I find them both fascinating and amusing, in more or less equal parts, and observing them play out, particularly on social media, is a thing I do for fun. I spent the entirety of the Fall semester trolling Donald Trump on Twitter so I could analyze him for my Capstone project. It was the most fun I’ve ever had doing homework, but I’m saying that to say this: 

We’ve been watching the 2016 Presidential Election Circus for a few months now, and I’d be hard-pressed to find a candidate that didn’t at least pay lip service to religious interests. For this election’s round of Republican candidates, religion seems to be a most necessary component. We have Ben Carson with his Seventh Day Adventism and Ted Cruz and Mike Huckabee with their Southern Baptism, and we have, of course, Donald Trump, whose views seem to change depending on who he’s talking to at the moment. The ‘Is Obama a Muslim’ situation and Mitt Romney’s Mormonism and Kim Davis checking out on her job and calling it standing up for her religious beliefs. So is every political dispute now a moral dispute as well? 

There is some question, of course, as to whether religion, or, more specifically God, is a moral issue at all, but if that’s the case then we may as well just throw in the towel now, because nobody can really win. 

Rosenberg suggests that moral disputes are intractable. Me being me, I looked that word up in the dictionary just to be sure. Intractable means hard to control or deal with, and in a person, it means stubborn. 

Seems about right. 

Teaching robots (or cars) to be ethical may be an easier job than teaching politicians to be ethical. Teaching a machine to be ethical has its complications, to be sure. But I suppose that as long as you don’t teach the machine to also have free will, you’ll have more success with it than you’d have with a person. 

That most people would applaud others’ use of a utilitarian car as long as they don’t have to use one themselves is telling. Once we have to start thinking about all the implications of this, our heads start to bubble their ways toward explosion. Or mine does, anyway. 

It isn’t just the autonomous car itself that’s the problem. It’s also how the autonomous car is used. Take for example the movie The Fifth Element. I’m sure there are other movies that depict the same sort of thing, but that one just happens to be one I was forced to watch because I’m married. I won’t attempt to get technical here because that would be painful to watch, but basically, the cars are all connected to one big system. The police can not only stop a car, they can stop a car and open the door. Now, you say, what’s wrong with that? It eliminates dangerous car chases, blah, blah, etc. The police pull you over, you stop. They want you to open the door, you open it. And I’m not advocating the evasion of arrest in any scenario, but don’t you like feeling like the choice is THERE? And what about scenarios in which the police are actually the enemy? Again, I’m only saying what if, but we know that’s been known to happen. Shouldn’t we have the option to drive to a well-lit, well-populated area? Shouldn’t we have the option to keep a door closed if we fear for our safety? 

It’s a conundrum. 

I suppose that's where Asimov's Three Laws of Robotics come into play. As a system, it does seem rather foolproof, but in regards to the discussion on machine-learning, there is that: They can learn. This can be a good thing and a bad thing, I guess, because it seems like if a machine can learn, then a machine can learn to break laws. Especially in the interest of self-preservation.