This massively reminds me of the "The machine fired me" [0] post from today morning's HN front page. Or, for what it matters, of the inhumanity that currently goes on in the US, with even toddlers being detained (https://www.huffingtonpost.com/entry/tender-age-shelters-tru...).
Following orders, eh?
> I can't fathom a situation where we want to automate what should be human interactions.
I can, it's called fascism. Break the individual responsibility of a human down and you can get them to do anything. There is a reason why in firing squads some members always have blank magazines [1] and why in the lethal injection sites a random of the required buttons for delivery will not work[2]. The individual member can try to evade responsibility by thinking "it was my button/gun that had the blanks" (apparently called "Diffusion of responsibility").
I recommend everyone able to do so a visit to Europe and one of the concentration camps the Nazis built. No matter which one, there are more than enough sites to remind us of what humans can do.
[0]: https://news.ycombinator.com/item?id=17350645 (thanks to @larkeith for pointing it out)
[1]: https://en.wikipedia.org/wiki/Execution_by_firing_squad#Blan...
[2]: https://en.wikipedia.org/wiki/Lethal_injection#Procedure_in_...
/s
Compared with certain death, either due to hunger or being forced to fight for drug cartels or in pointless wars, fleeing without a safe plan is the lesser evil.
I can. When you're a for-profit organization. People are expensive.
This is basically late-stage capitalism.
Really? You'd rather call the taxi company and make an appointment with a person than use Uber? You'd rather ask a librarian to do your google searches for you?
Citation needed. You're neglecting the fact that there are millions of under-interacted kids in the world. Providing a cheap source of interaction for them could be extremely valuable. Just saying "their parents/teachers/whoever should do it" does absolutely nothing to solve the problem. Building them a robot does.
> "should be human interactions" is a meaningless standard
What irony! The entire concept is that there is a place for real human judgments and not just rigid programs/rules!
What "should be human interactions" is not meaningless. It means: there should be humans judging what should be left to human judgment. It's an assertion that there ARE things that shouldn't be just standards we blindly follow.
It wasn't saying any human interaction automatically matters. Tasks where a human would just follow rigid protocol are ones that should be automated…
You mean...humans like the humans who decided to buy this software? Presumably you don't just mean 'humans'. You mean the 'right' humans. But which humans are those? Who decides which ones the right humans are? It's just a circular argument that reduces to "humans should do human stuff", unless you want to actually articulate a standard, but you have definitely not done that yet.
Bad presumption. The argument is about the difference between humans versus computer programs. Whichever case, I want the 'right' ones (i.e. the best ones doing good not evil), so the 'right' ones isn't the differentiation here.
Whatever the argument is, there's an argument that having real people instead of programs matters for some things. And the standard we can use is something like "the things where they aren't interchangeable" at least (although others may have other qualities to emphasize). So, a person who literally just does the exact same thing a program can do can be replaced by the program. For example, data entry. A person who cannot be duplicated nearly exactly by a program is where we want to be careful about considering keeping the human position around. For example, teachers.
I heard a quip that computers are passing the Turing Test more these days… but not just because they are getting more sophisticated — human beings are failing the test more (are acting more like computer programs). The point being that "human interactions" in the argument here implies humans engaged in natural human thought and intuitions and communication and not just following protocol so rigidly that they effectively are just carrying out program instructions.
I agree. But my point is that most people who talk about these things don't do a very good job in thinking about which things should and should not be automated. They basically just look at whatever has been automated (which they love being automated, now that it has been) and say "Automating that stuff was great!", and look at the things that haven't been automated and say "Automating that stuff is bad! It is essential to the human experience!".
> For example, data entry. A person who cannot be duplicated nearly exactly by a program is where we want to be careful about considering keeping the human position around. For example, teachers.
Why? Why would automating teachers be bad? I'll give you two separate arguments why this would not be a bad thing:
1. Access: Teachers require human labor, which is in finite supply. This means that not every child on earth has access to teachers, and sometimes if they do, the teachers they do have access to are spread incredibly thin, making them effectively useless. Automating teaching has the potential to massively expand access to the extremely underprivileged. Teachers also operate on a set schedule. If the class starts at 8pm, and you have a job that requires you to be at work at that time, well, tough luck for you. If you automate teaching, this problem goes away.
2. Personalization, optimization, and equity. Human teachers can tailor their lessons a little bit to the individual, but high quality AI instruction promises to expand that greatly. Also, not all human teachers are good. Figuring out which ones are and are not good takes a lot of time, and wastes thousands of hours of child development, and sometimes doesn't happen at all. If we had artificial instructors, they would be the same for everyone. We would be able to optimize their methodology, and ensure they are teaching at a minimum standard of quality - however we want to define that. This also opens up massive opportunity for improving equity among races, classes and whatever other social divisions you want to impose. Now, that isn't a given, AI has the potential to encode our own biases. But unlike in human instruction, we actually have the opportunity to correct those biases if we're careful in designing the AI - we have no such opportunity for human instructors.
The problem, I think, is that when people think about AI replacing teachers, they think about all their favorite experiences with human teachers, and how those might be lost. And that may well be true. But what about all of the negative ones? What about all the kids who's teachers were lazy or having a bad day that day? AI doesn't have bad days. It doesn't get tired. It doesn't dislike children because they're queer. It doesn't get frustrated by insolence. It teaches at the same standard of quality every day.
So yes, maybe some of the great human experiences may be lost. But so too will the negative ones. And it's really not obvious which side of that equation is more important. But that's sort of the point: it's not obvious.
I mean, everything I'm saying is based entirely on taking a charitable interpretation of someone else's comment. Instead of asserting that a comment is just a useless circular logic, I'm saying that we can make sense of it better.
An assertion like "should be human interactions" is an assertion that there exist things that will never be better automated (at least compared to the higher-quality examples of human interactions). I'm disagreeing with your argument that such an assertion needs to specify which interactions those are. It's merely stating that it's a non-zero list and leaving the determination to others at a future time.
It points out that there are many things which are automatable, but that the capability to automate things cannot be the decision driver of whether things should be automated.
At one end of the scale, ATMs replacing some bank tellers seems to have been widely accepted as OK. In the middle, self serve supermarket checkouts (here in .au at least) are saving staff costs for the supermarkets with the side effect of pushing shoplifting prevention and enforcement costs off to the local police instead of supermarket staff, which is not such a great automation outcome (unless you're the manager who got the bonus for reducing headcount costs at the supermarket). At the end alluded to in the discussion of this article, I think it's pretty clear that attempts to automate discipline of children is an abhorrent idea.
Not the OP, but yes, I actually do that, I mean, I've avoided installing an extra app that would track my every move (the Uber app) and I prefer calling the taxi company on the phone. I also very frequently take random taxis from street corners or the like.
"Are you kidding? Too many people write things down, it causes forgetfulness." - Socrates
I don't see what value a 2 minute conversation with a human who is essentially following a script and pressing the same buttons on their end that you could have pushed on yours has. Despite the audio portion of the human's voice, this is still an extremely shallow form of human interaction. You are better off saving the time and having a more meaningful connection with someone else later.
I find this attitude extremely alienating, at least in my case. I've been through several dark periods in my life and having constant human interactions even for "menial things" was one of the things that has kept me above the water, so to speak. I also don't associate human interactions (even with strangers) to making "transactions" or the like, that, to a person like me, looks like treating people as means instead of ends in themselves (I'm a kantian when it comes to morals, or at least I try to). We cannot keep a compact and well-functioning society if we start seeing talking with other people as "menial things" or "transactions".
> as far as CC details, we have fraud protection
In my country (in Eastern-Europe) CC fraud protection is a joke. Even if that weren't the case, I don't want a 3rd party to have access to my finances if I can avoid it.
Also, yes! I would ask a librarian to search for me! Librarians are AWESOME and can find things in places you didn't think to look. Everyone should ask a librarian in addition to doing their own Google search.
I felt unsafe every damn day in high school, and it certainly wasn't because some facial recognition tech was watching my every move.
But whatever, keep trusting those kids to eventually behave. Personally, I just hope karma got them in the end.
I was homeless for nearly six years. I got off the street last September.
I currently live in a dirt cheap SRO. I have lived in high crime neighborhoods and was last in Fresno while homeless (where murder and arson were routinely in the news).
"Eyes on the street" is how you prevent such things. Human monitors who can serve as witnesses are a deterrent to crime in a way that cameras are not. Furthermore, adult supervision and the right policies can help troubled kids learn how more pro social behavior.
I'm talking about prevention, not sweeping it under the rug and ignoring it.
Agree! Could you imagine having millions or billions of records of students learning the same set of concepts? Big data pedagogy, here we come!
> Kids skipping class, kids beating up other kids, even someone walking in with a weapon
Oh... Disagree :-(
Schools are places of learning, not prisons. Technology in schools should enable learning, not surveillance.