Now, interesting question: given that editorial columns in prestigious publications like the NYT exist as a part of their own pecking order of columnists, is Douthat any more perceptive than some person on medium.com? I think this article had some interesting things to say (even if it happened to be long-winded), but some of the comments on the NYT disagree that the Times was so right to feature him.

Also, if I am right about so much about Reid's arguments about how scientists should go about advocating for science really were secretly about the form and emphasis their statements acquired once published in a particular publication, then all I can say is that we ought to have simply heeded Marshall McLuhan: the medium is the message! (Uh, not the website medium.com. no pun intended, lol)

The fact that Douthat's opinions are published on the NYT doesn't mean that they're intrinsically superior or more perceptive than a medium user. But being associated with the NYT gives him prominence, guarantees a large readership (and also that important people will read him), and gives him an air of legitimacy as a voice that is worth listening to.

RE: the Douthat article. I'm kind of blown away by the cynicism of the center-left going after Facebook. I'm completely in favor of the public conversation we're having about Facebook, but can you imagine it happening if HRC won the election and Sheryl Sandberg was the first female Secretary of the Treasury? lol. Plus, I'd love to see how the Clinton campaign acquired its data.

Reply to your first comment: Yes, that's more or less my view. It's just that this article was ****ing long, heh

I read your comment again and thought I might've misinterpreted what you were saying.

Oh, I agree with the thrust of Douthat's article. And you are right about how pathetic the center left is on seizing on this Facebook stuff.

Because what you’ve suggested isn’t the same thing as what you claim you’re suggesting:

‘Stephen Hawking shouldn’t have expressed his own opinion about AI, he should have cited the opinions of AI experts’

disregarding the fact that I’m not sure who is qualified to be an expert on strong AI other than science fiction authors and insufferable transhumanist blowhards, I’m also not sure how you can consider an evidence-free and weakly argued statement of some dudes’ opinions to be scientific research.

What you’re actually saying, as evidenced by your own example, is that you want people to “take academics’ word for it”. And like you literally said, you want advocates to teach people how to recognize who is an expert and deserves attention (because they are accredited by the academy).

I don't think you did, to the extent to which I was saying much at all. At any rate I agree with you.

Here's the NR article he cites. It's even more scathing in tone. And it also discusses Obama's tactics for gaining data of its users. It was almost identical to Cambridge Analytica. lol https://www.nationalreview.com/2018/03/cambridge-analytica-social-media-panic/

Haha, I remember when Democratic supporters were gushing in the years following Obama's victory. They were particularly proud that his team had so expertly gauged every electoral battle, and had used this information to pull off a victory. Lol

I find academic arguments about strong AI exhausting, regardless from which direction they’re argued.

The fear mongers (to generalize their arguments) are afraid that strong AI will go postal, exterminate or enslave humanity and single-mindedly turn the universe into forks or something. Human-like intelligence isn’t a theoretical concept though, it exists, and almost 100% of known human-like intelligences aren’t at all like what these experts fear. Do humans have problems and deviances? Sure, but we’re also capable of policing that behaviour. Could human-like machine intelligence be very different from bona fide human intelligence? Sure, but we’re doing science here. If you’re proposing that theoretical strong AIs will necessarily be antisocial, you’re contradicting exceptionally strong evidence that (inter-species!) social behaviour is selected for intelligent and conscious animals. These same scientists have yet to provide a satisfactory explanation for why humans aren’t, every one, a lying, Randian rape-monster, so I’m not sure why we should trust them when they speculate the same about eventual strong AI.

Same thing for the transhumanists (and others who argue or fear that strong AI will beget stronger AI, until approaching ‘infinite technology’). We already have an example of an intelligent, conscious machine that is set upon improving itself. What makes you think humans aren’t already improving their intelligence as fast as physically possible? We don’t even know what consciousness is or how it works, what structures enable consciousness to happen. How do they know that the human brain isn’t already the most optimal and compact structure for generating consciousness? The human brain clearly isn’t perfect - computers have more working memory than we do, and are much better at computation. But the fact that computers are better than brains at computing doesn’t mean they will ever be better than brains at braining. Is it possible? Sure, but you have no evidence for it. We’re supposed to be doing science.

It’s all a bunch of baseless speculation. Everything. The only thing we can say for sure is that Google will misuse AI, but that’s because they misuse everything.

And trying it back to politics, I’m extra chuffed that most of the arguments against strong AI also apply to capitalism. An unthinking machine that enslaves humanity and single-mindedly converts all resources into a useless product? Hmmm.

Eh, the question about AI's benevolence/malevolence seems so naive to me. I'd be surprised if AI academics talk about these questions all that much. It almost seems to assume there will be some one moment when suddenly AI bursts onto the scene, and when it does, it'll either be benevolent or malevolent. It smacks of sci-fi understands of AI that make the emergence of AI will be a dramatic rupture, similar to the sudden moment when we make contact with aliens and the suddenly enter into human history from without (never mind that it sounds very messianic). It seems much more likely that AI's emergence will be much more gradual/incremental, and that if AI has any particular characteristic, it only has it because humans decided it should have it (and had to work really hard to figure out how to make it work).

That doesn't mean that killer robots aren't a real problem. But killer robots can exist without developing anything that approaches human "consciousness".

I take it then that you don't buy into the idea of a singularity then.

To me weak AI is just another synonym for automation that isn't being scrutinized by humans for potential flaws.

Strong AI is basically like a human, which thinks on smaller time scales, so that we don't have time to react to it, and can generally thwart our attempts to control it. Of course this is less of a problem for autonomous beings, since as Jon`C has mentioned before, these more or less converge to existing human beings (and would such a sentient being be much more effective than a human who knew how to use "dumb" computers to augment their own intelligence?). What would be more concerning is if humans deferred completely and universally to some machine intelligence that used electronic communications networks to simultaneously replace human intelligence across a wide cross-section of the economy. Of course this is literally what automation is, and I would be surprised if such automation became completely autonomous to the point of resembling a sentient being, before it automated away most of the jobs in the economy. And this would be especially interesting, since as Jon`C said, the arguments against AI become synonymous with those against capitalism.

Of course I doubt that AI would want to kill humans, any more than humans feel the need to kill livestock. Unless you are going to argue that they need to eat our brains for energy.

I mean, as Jon`C remarked before about capitalism before, we're basically already living in a sort of open pasture prison (edit: couldn't find these words by searching, so maybe he put it differently) anyway, so I doubt the introduction of AI would change that (edit2: although are dating apps a precursor to a sort of machine managed "human husbandry"??). About the only way that could happen is if the AI realized that humanity were a threat to its own survival (which I believe was the plot of at least one Hollywood movie, based on a well known science fiction story).

“Free range slaves”, or something equally bombastic.

OTOH, rather than bracing for the prospect of being globally enslaved by machines programmed to optimize for efficiency according to selfish, unfeeling and merciless rules, what about objectivists?

objectivist function

Another thing about this in terms of some of Reid's points about experts and his other prescriptions about public discourse. One of Reid's implicit desiderata is that the public should have access to certain facts. That is, it's not enough for the public to know true facts, but they should be certain that they are true (which is to say they should know why they are true). But this fundamentally isn't how the relationship between the press and the public works. The media doesn't offer certain facts, and it's goal isn't to provide certainty. Our relation to the media depends fundamentally on trust. We aren't in a position to figure out for ourselves whether the reports we read in the paper are true (or at least we rarely are). We can only believe that the outlet that's providing us with information is, for the most part, reliable and trustworthy. The authority of academic experts is very different from the authority of journalists and other experts, in that regard.

So that's a crucial difference between a NYT columnist and Medium. As an institution, we have no real reason to trust Medium, because it's just a platform that literally anyone can use. But we should be able to trust what we read in the NYT, because everything it publishes is backed by its reputation. Of course, that doesn't mean that the paper always deserves our trust.

lol'd

is this the thread to discuss uber finally killing someone, directly? i got real angry about that earlier today when someone tried to blame the woman for jaywalking and getting herself killed.

You should get real angry. A lot of people online have been claiming that she erratically jumped out in front of the car (and claimed that the video showed it). I assume Uber's been paying people to say this, because it's not what the video shows. The video shows that the car failed to detect a woman who was normally walking a bicycle across a road. The video shows the driver, who was supposed to be supervising the car, wasn't paying attention. Because like we've known from the past 40 years of automation in aviation, that kind of automation makes drivers dangerously complacent and incapable of responding to emergencies.

i hadn't watched the this morning, but i watched the video clip just now. i wasn't able to see the pedestrian until right before the collision, however there was plenty of time to slow down/stop if you were using a ****ing LiDAR system to detect the surroundings. now i'm even angrier.

this is what i get for trying to say that a current trend in tech is irresponsible to a bunch of mid-20s tech workers.

I'm not embedding but this is the video: https://youtu.be/XtTB8hTgHbM

It cuts off before the collision. Watch the second half for a rad video of a dude looking at his phone instead of what his car is doing.


Is Uber optical like Tesla, or are they still using the LIDAR stuff they stole from Google? If they're using optical it might explain things, since she was wearing dark clothes in a poorly lit area without reflectors. Not that reflectors would have necessarily saved her, depending on whether the car could recognize them.

If they're using LIDAR there is absolutely no ****ing excuse, and the cars need to be taken off the road immediately.

Not to blame the victim, but this is a good object lesson about jaywalking and using proper bicycle safety equipment. You never know when some deep pocketed corporation is going to kill you and blame you for it.

move fast and break things

If they're using optical sensing and optical sensing can't even tell the car to slow down just because the object was partially obscured, then in my opinion the technology is not sufficiently safe to be allowed on the road. A human would have had plenty of time to react to this.

What's important is that we aren't letting engineering ethics get in the way of innovative Mom as a Service companies.

https://www.nytimes.com/interactive/2018/03/20/us/self-driving-uber-pedestrian-killed.html this says LIDAR

We can safely trust that is true, being from nyt

You are supposed to look both ways before you cross the road...

Legal question: are there laws against running over j-walkers? My understanding was that if you are able to prevent an accident (of any kind), you are expected to at least try to yield, even if you have the right of way.

Varies by jurisdiction, but the standard is comparative fault. Case by case. I believe things like distracted driving do count against you, even if the accident couldn't have been reasonably avoided anyway, but IANAL.

We don't know if he was looking at his phone and not the car telemetry.

The cars have pretty good stats vs. human drivers, probably spared some lives from human error so far by driving those miles autonomously. I'd wait for more information on the incident before taking the cars off the road.

It's pretty difficult to appreciate who John Bolton really is when people are literally saying "it's time to panic now" and "we're all going to die". Like most articles about him, this one doesn't contain much description about why he's such a dangerous choice, even as it incorrectly labels him a neocon.

since 2016 when was it not time to panic?