In a community with so many members that pride themselves on contrarian views, why is this view on autonomous vehicles and more broadly, incredibly powerful AI that is just around the corner so broadly accepted?
Not sure about AI since I am still skeptical about "incredibly powerful" AI around the corner. AI is definitely going to have a big impact in the not too distance future but my sense is that the AI need not be all that powerful, it just needs to be capable of eliminating human input.
Self driving cars on the other hand are right in front of our eyes, we have seen them driving around and are rapidly advancing in terms of their level of autonomy...in terms of timeline for widespread adoption, I'll look at utility. The utility of self-driving cars is just too great for them not to quickly replace manual cars once the technology is fully ready. I can't imagine too many people would want to be stuck fiddling with their cars in long commutes when they could be reading, watching TV or taking a nap...just as smart phones dominate now because of their utility, self-driving cars are likely to unfold the same way.
>Self driving cars on the other hand are right in front of our eyes, we have seen them driving around and are rapidly advancing in terms of their level of autonomy
There's basically no evidence for this though. Sure Tesla can show me a 3 min video of a car driving in perfect weather on perfect, straight, wide roads but I don't see self-driving cars "in front of my eyes" at all. We don't have any safety data that is directly comparable to human driving, we have several accidents already, etc.
>when they could be reading, watching TV or taking a nap...
I agree that when/if we make it to that level of autonomy we'll have turned the corner on adoption and it will explode. But I'm not sure that even the most optimistic of self-driving companies would say that we are anywhere close to that.
You'd have to wonder though if this is perhaps scoffing at the notion of an airline industry at the time the Wright brothers are showing off their first aircraft.
Getting from "man powered craft that can stay in the air for 30 seconds and a hundred odd metres" to "jetting millions of people around the world every day" requires a leap of imagination, but unequivocally it was possible, and happened.
Like the uncanny valley, the edge cases for self-driving cars will take more work than the rest of the effort combined... by at least an order of magnitude.
There is far more money in automating long-distance trucking. My bet is the first real use of self-driving "cars" will be for trucks. The human driver can sleep, eat a meal, etc but will remain on-board for safety's sake and dealing with unanticipated edge cases (e.g. the AI wakes up the driver when it runs into a traffic jam due to an accident. The AI can probably recognize cop cars, flares, etc but won't be able to navigate the scene of the accident while responding to instructions from police officers)
The next small step will be "follower" trucks. The lead truck has a human driver same as before, but one or two AI-driven trucks follow. They do their best to track the human driver's response to unexpected situations. Worst-case, they pull over and flag the human driver who pulls over and uses a VR rig to pilot them past the problem.
Meanwhile the application of "self-driving" technology to cars over the next 10 years will be to avoid accidents, not to actually be self-driving.
Will we have fully autonomous vehicles within 50 years? Almost certainly. Will we have them in the next 5? Almost certainly not.
Self driving haul trucks are already at work at a few open pit mine sites around the world. These are vehicles driving relatively short, prescribed routes, with well defined patterns. The few articles I've read seem to indicate they're operating just fine, although I'd be interested to learn more. If the cost savings are there, I'd bet the entire mining industry will move in that direction rather quickly.
I'd argue that the shit self driving cars we have right now are ALREADY statistically safer than humans.
You don't have to be perfect. Just better than an average driver. And average drivers are drunk, sleep deprived, and texting. That is a very low bar to pass.
To sell you need the car to be a better driver than the customer thinks they are which is a higher bar since most drivers think they are better than average.
Maybe on highways in good weather conditions we are close, but the chaos of cities like Paris, Madrid, Rome with their narrow roads and millions of edge case situations will be a tough problems for self driving AIs.
A number of us live in Silicon Valley or work daily with AI. These aren't science fiction technologies for us: they're, real, tangible cars that you see daily on the roads, or software libraries that you can download and use yourself. I can understand why they'd seem like fantasy if you live in some far-off place, but when you're stuck behind a self-driving car going 25 mph on El Camino, there's pretty concrete evidence that self-driving cars are here.
I do think that AI and particularly deep learning are overhyped. It is a big deal, but it's not as big a deal as many people make it out to be. It's much like the Internet in 1998; it works, you can use it now, and you know that it's going to drastically increase the reach of computing power, but people are talking like that's all going to happen in 1-2 years when really it'll be 2-3 decades.
I guess my concern isn't that these are "science fiction technologies" but that they're rather simple. AI has a history of over-promising and under-delivering and even if a few "self-driving" cars have made it on the road, how long until people begin trying to use them in scenarios they're not really ready for? All of that progress can disappear overnight if there are a few bad accidents.
I guess when I look around, I don't see a comparison between AI now and the Internet then, I see a comparison between AI now and AI then (70's).
It's worth remembering that many of the actual technologies developed in 1970s AI - A*, beam search, pattern matching, constraint solving, rules engines - are now basic CS algorithms that most programmers are just expected to know, and they have continued applicability to a wide variety of domains. We just don't call them "AI" anymore, because the technology has been commoditized enough that they are "business rules", or "single-player games", or "layout engines", or other examples of the actual problem domain that's solved.
Ditto 1990s AI - fuzzy logic, SVMs, Bayesian networks, collaborative filtering, perceptrons. It just got good enough that it became the standard way you solve a certain class of problems with computers, and now it's called "fraud detection", "Zestimates (tm)", "recommendation engines", "stream ranking", and so on.
Deep learning is the same. The hype cycle will eventually bust, because all hype cycles eventually bust. But in the meantime, it has opened up whole classes of problems in computer vision, speech recognition, recommendation, machine translation, and several other domains that were not amenable to computerized treatment before.
I guess that's sort of my point though isn't it? A, SVMs, etc. aren't really called AI because they're not. Because AI carries a connotation and once you understand A it's obvious that it's not really AI. Won't the same be true of today's neural networks? Granted they've been given a name which should add some longevity but when a layperson thinks "AI" they think a robot which acts like a human, they think AGI. If someone had an AI algorithm which could start to demonstrate more AGI leanings I bet the AI effect wouldn't apply.
When cars came out they killed quite a number of people in accidents and mechanical failures. They were just so much better than horses we kept on improving them.
That's how I see AI. People will surely get injures and even die in Accidents. We are a humanist society where human life is valued above everything else, so I'm very optimistic that AI will get better and give their humans more divine powers and longer life over time.
Even assuming your conservative adoption/relevance curve of the technology itself, I don't think we need too many more incremental steps before society has to stand up and take notice of an unemployment problem that can't keep up. AI / automation might not turn our day to day lives into the Jetsons in 5 years, but if it gives us 25%+ real unemployment we're still going to be in for a ride.
Traffic in the US also seems vastly easier than in other parts of the world, mostly because roads are wider. When i see the some of the chaos in cities like Paris, Barcelona, Rome and the narrow roads in many places, i can only come to the conclusion that it will take a long time before an AI will be able to handle that. Highway/intercity traffic sure, that should be solved in a few years, but the rest will take some time.
Even in Germany we have lots of backroads on the countryside that are so narrow (and have no markings), that you have to go off the street with half of the car if someone is coming from the other direction.
Who pride themselves on contrarian views? That looks like a rather foolish kind of goal.
Well researched, fundamented, empirically valid views are worth pursuing, and if those happen to lead into contrarian, ok, nothing wrong with that, but pursuing it? No way.
Or of course, things could take an altogether different turn. I remember when GPS was introduced in 96 for civilians and no one could have predicted Uber at that time, though all the pieces were there. BE does a great job of synthesizing possible directions from what's available to him in albeit a really good position at a16z. However, there are many people out there quietly building technology that changes the game completely. Making affordable graphene is one such unknown at this point.
Another example of having the technology in front of you and not predicting its use is the Apple videos under Sculley. I remember one (cannot find now) that had a man scan in a newspaper into his tablet to help him learn how to read.
On the other hand, in 2001 (eight years before Uber):
Basically WebTaxi works well with people having wireless web access from PDAs (Personal Digital Assistants). Imagine a scenario where you are sitting in a restaurant and you need a Taxi service. You go to the WebTaxi website and request a Taxi to your home. WebTaxi comes back and tells you exactly how much it will cost you. It can even tell you how long you will have to wait for the taxi. You continue to chat while you wait for the taxi to show up. As soon as the taxi arrives, your PDA informs you that the taxi is waiting for you!
I think the rise of machine learning, graphene, high performance electric engines and metallic hydrogen will finally allow a reusable vehicle to go into space, something like the Skylon [1].
The reason why I bring this up, as being important in 10 years or less. Is that it will be available for around $1m.
AI will negate the need to go through rigorous astronaut training. Maybe there will be some, but not as thorough.
Graphene and other exotic metals will allow for lighter craft and reusability.
High performance electric engines will allow for less fuel payloads to get into high altitude, whilst allowing for the metallic hydrogen to literally blast out into space. Metallic hydrogen is 7x more powerful than liquid hydrogen.
I think once average joe is able to get into space. We could finally see an abundance of rare earth materials. Could be like the old west, where a co-op of you and your buddies go out, wrangle an asteroid and mine. Come back with a trillion dollars worth of metals.
Of course, a glut of anything will bring prices down. But hopefully by then manufacturing is ubiquitous also.
Maybe it's the optimist in me, but there are certain historical events that could take our society from where it is now, scarcity based, to an abundant one.
Graphene is interesting I was working on synthesizing it for my undergrad thesis in Materials Engineering in 2010 (back then we were looking at using it in Li-Ion batteries).
Personally I feel we are still more than 10 years away from mass applications. I'm happy to be proven wrong though.
Thinking from a Materials POV in the next 10 years I think 3D printing (additive manufacturing) is an area where large advances will occur. New applications for 3D printed materials and wider adoption of 3d printing.
Considering how much more abundant we are, in terms of CPU, RAM, Disk space, compared to twenty years ago, and also in terms of food, clothing, shelter, compared to a thousand years ago, IMO we're only going to find more ways to consume the abundance in ever more less efficient ways, and won't escape "scarcity" within the next 100 years.
"mobile is maturing and its growth is slowing"...yes, somewhat on the hardware side. There are still annual sales of maybe 1 billion smartphones a year...so that's a pretty big market.
I think that there's still ENORMOUS potential for mobile apps and mobile-driven services. There's huge opportunity for new services e.g. Uber. I think just about every large web/desktop property faces a significant threat from a mobile app e.g. craigslist, match.com (they luckily own tinder), flickr (wait, already gone to Instragam), AIM/Yahoo Messenger (gone thanks to Whatsapp), etc...
True, but the pace of the transition was substantially slower. I think that is the concerning part: it will happen over years (less than ten) versus generations.
> Or they'll do better jobs. [...] The 69% didn't become an unemployed underclass.
No, they moved into factory, mining, retail, and other jobs that were commensurate with their abilities. The automation shift is not the same, because it will take over all the unskilled/low-skilled jobs, and there will be nowhere in the economy for the displaced people to move into.
It is not really conceivable that all these people are capable of taking on jobs which require complex thinking, judgment, creativity, or other higher skills. IQ is a real thing, and although it does not reflect on someone's human value, it does very definitely place limits on what functions a significant portion of the population can perform. And in the long run, we will all be in that displaced portion. Best that we find ways to create human security and meaning, now, for a coming world in which we have no necessary economic role to play.
>The automation shift is not the same, because it will take over all the unskilled/low-skilled jobs, and there will be nowhere in the economy for the displaced people to move into. It is not really conceivable that all these people are capable of taking on jobs which require complex thinking, judgment, creativity, or other higher skills.
There are plenty of "low-skill" jobs that nonetheless require human labor in some form because of the ability of humans -- even less intellectual humans -- to think creatively and contort themselves both mentally and physically.
Imagine a robot that does everything a school janitor can do. That simply won't exist in the near future. Sure, maybe the janitor won't have to mop large floors anymore. But there's plenty of other work to be done around a school nearly always to continue to justify their employment.
I suppose it depends on what you think of as "near future", but I - and I have no special insight, for the record, just speaking as an outside observer - would put around 75% probability of a robot that can wholly replace a janitor existing in 15 years, and 95% in 20.
In terms of time needed to politically and socially prepare for that situation, 20 years feels really near term to me. I don't really expect our existing political structures and processes are capable of dealing with the challenge adequately in time. It's one of my top three fears for the <30 year future. And unfortunately the other two (global displacements from climate change, and bio-/nuclear-terrorism by desperate or angry people/groups) are synergistic with it. If it weren't for my normalcy bias, I'd be kind of terrified.
> No, they moved into factory, mining, retail, and other jobs that were commensurate with their abilities. The automation shift is not the same, because it will take over all the unskilled/low-skilled jobs, and there will be nowhere in the economy for the displaced people to move into.
I think a lot of people will be surprised at exactly which jobs too. It's not going to be the cleaners, the carpenters or the plumbers who still have to get into tight spaces that a robot butler can't handle. It's going to be the office workers that push computer bits around all day.
I think a lot of people will be surprised at exactly which jobs too. It's not going to be the cleaners, the carpenters or the plumbers who still have to get into tight spaces that a robot butler can't handle. It's going to be the office workers that push computer bits around all day.
Any job where it all goes in and out over a wire is vulnerable. But robot manipulation in unstructured situations is still very hard, which means people are needed for their hands. I've been calling this future "Machines should think. People should work" for years. It's getting closer.
I realize mentioning Bitcoin in a tech forum feels gauche in 2017...
Having said that, it's strange that this guy doesn't mention cryptocurrency anywhere in his post.
He spends a lot of time discussing technologies that aren't quite there yet, and leaves out the one area that is (likely) right at the point of adoption.
I was thinking the same thing while reading through this essay. I know the firm he works for (A16Z) is investing in the cryptocurrency area and even just did a podcast[0] on it.
Regardless of his opinion, he probably should have added a one-liner as to why it he didn't think it relevant.
Otherwise he leaves the reader wondering did he forget? or not think it will have impact? or feel it's outside the scope of what he''s discussing?
Regardless, it was an interesting read.
Agreed. I was mainly thinking about timing, but I let the words pick my meaning a bit there. I ought to have said that if bitcoin amounts to anything (ie: if new technologies built on it begin to affect everyday life), it's likely to happen over the next one or two years.
I think he missed the point about TV ad budget not moving to the likes of Netflix or Amazon. The reason it hasn't moved is not because the audience is limited but because the companies in question don't want to pollute their viewers' experience with Ads. Compare that to YouTube which is free and does have Ads.
Viewing habits are shifting but so is public opinion on Ads. The future of Ads on these internet TV sites is going to be the same as it was for traditional TV albeit a bit more targeted. Instead of TV networks controlling the viewing data in the future internet TV services will control it.
so the potential of a startup to take advantage of that data is going to be minimal since it will be guarded like gold.
(1) Software is well on the way to an old, slowly changing, mostly routine, boring, low margin, slow growth business and industry.
(2) That the general purpose computing hard/software and 100 Mbps Internet data rates are all so cheap will in places undercut large, expensive approaches to computing projects. E.g., people will do projects, but Sand Hill Road need not be more involved than they are now in, say, a new pizza shop in Peoria, not more involved than they were in the romantic matchmaking Web site Plenty of Fish.
(3) A huge problem for computing causing (1) is that there just are not enough good, new ideas for high value IT startups. For a joke, but a simple explanation, the days of a hot social, mobile, local sharing app are gone.
(4) For Evans's dreams about machine learning (ML), he is not very well informed and is not seeing very clearly, but in principle, if strain to extend the meaning of ML, he is correct about the potential. But in practice that future will come slowly, very slowly.
Why so slowly? In fact, e.g., looking at, say, scikit, that's basically some applied math, and nearly all the math there is now old enough to have long, gray whiskers, literally. That applied math and much more in optimization, simulation, experimental design and analysis of variance, stochastic processes, on and on, i.e., the math sciences, optimization, operations research, statistics, except for a few niches (e.g., optimization in operating oil refineries, the fast Fourier transform for oil prospecting, partial differential equations for modeling oil reservoirs, air line crew scheduling, maybe airline plane scheduling, statistics in biomedical research and development) hasn't caught on at all well generally in business; so there is only tiny reason to believe it will become "The next big thing" now.
For that math, it's not so easy to get good at it: (A) The pure math departments don't want to teach it. (B) Very few other departments have enough grounding in the math prerequisites to teach the material very well. (C) Except for a few niches, business has long made its position crystal clear -- such applied math and a dime won't cover a 10 cent cup of coffee. (D) It has long been the case and still is that about the only paying customer that takes such applied math seriously is the US DoD for purposes of US national security.
Run a good applied math project down Sand Hill Road, for a big problem, with running prototype software, and will hear laughs, silence, or "don't call us; we'll call you". I have to doubt if there is a single information technology (biotechnology can be different) VC firm on Sand Hill Road able to direct technical due diligence for an applied math project.
There can't be "The next big thing" without a few hundred times more well qualified startup founder, worker bee, applied mathematicians to do the work.
Excellent questions regarding autonomous vehicles. Instead of a city center being two hours away at rush hour, it could be a half hour trip.
But also living in the city is going to get a lot easier because of robotic transportation. It won't be necessary to haul anything home from a store, for example. A lot of cabs could just have one or two seats and no trunk.
We'll see tiny vehicles delivering dinner, handling dry cleaning. And parking woes disappear.
But the biggest change will be in how easy it will be to move from one place to another. Yes, robots will be able to move all your stuff cheaply. But they also might just procure comparable stuff in your new city instead. And most of your stuff you don't use all the time, so maybe it's in long term remote/cheap storage until you do. Maybe its okay if those table leafs are at your house for Thanksgiving dinner.
Maybe you move every month or live in one place during the week and another on the weekends.
This sounds good in theory because it correlates with trends in the IT industry. Centralized compute/storage/app delivery on an IaaS but in the real world, with our actual physical stuff instead of data. In practice I'd wager that this lifestyle of moving constantly and having robots move your stuff around for you would be prohibitively expensive. Also, who wants to containerize their life? How could this ever be sustainable for someone with kids.
I think we should be less focused on autonomous driving, and more focused on reforming our tax code, infrastructure spend (too little), and healthcare spend (too much).
I know a guy that travels constantly and has no permanent address. And there are a few families that spend years on the road because they want to teach their kids that life is an adventure.
They haven't found it prohibitively expensive or unsustainable. It's worth it to them. Making that lifestyle cheaper and easier will expand the number of people for whom that is true.
Automation, as it has always done, will make our infrastructure cheaper and more robust. It will slash the cost of healthcare and dramatically increase the amount of resources we have available and our quality of life.
So, no, I don't think we should focus more on nibbling around the edges of what is possible. Those things can all be fixed, but that takes political will and convincing the majority of people. And most of them are stuck in the analogies of the past.
Robotic transportation could also greatly increase the demand for cars and actually increase traffic on the roads, even though less total cars would exist because they would be better utilised. Even public transport could disappear if robotic cars could meet demand, if they can't, they will probably stay more expensive than public transport.
Higher ed should see the most impressive shift of all in the next 10 years. Right now, it seems like everyone realizes there is a problem, but nobody can see the future.
Student loans are driven by massive inefficiencies that various companies will exploit. The end will be apparent in 10 years.
For the US thats true, as a german who can basically study for free at any of the "elite" Universities in the country (same for scandinavia, netherlands etc) the whole higher ed concept in the US seems borderline scam to me anyway.
The German "elite" label does not mean anything at all. It's basically a government program to allocate (a small amount of) additional funds decorated with a fancy term.
Well yes, but all of them are among the most highly regarded Universities in the country, while there are others without the label that are equally good. I just wanted to emphasise that the quality you can get for "free" is really good.
People can't really see it yet, partially due to a unwillingness to see the darker side of humanity, but a real cyber security expert is worth more than a cruise missile.
Planes, self-driving cars, swarms of aerial and underwater drones, fraud, propaganda, auto-targeting laser systems, surveillance, power generation and distribution, command and control interruption, machine learning + cyber warfare.
Who needs nuclear weapons when you have a zero day on a hundred million self-driving cars and you can get them to the nearest gas station or chemical plant going 300km/h?
I wish the present was a little less obsessed with having everything Internet-connected. I don't want an Internet connection on my self-driving car, or at least have an air gap between the user functions and the car's driving computer. But that won't happen because the company will be more worried about being able to push updates than about security.
To be fair I guess even with a non-connected car that I have to update via USB, I could still download a malicious update. But at least we won't all get the malicious update pushed at once.
All I see is praise for Tesla on HN. I rarely see sentiments like "Tesla makes great cars, but I just wish their ECU and other critical systems wasn't internet connected" or "I refuse to buy any Tesla products until they remove the constant user tracking".
People vote with their wallets, and if the technology-savvy and privacy-conscious HN crowd is voting this way then...
.. nothing changes. We are too small of a group, even as influencers to change that.
However, if someday a zeroday is used to duke it out, a very angry, very unreasonable, not interested PR-Bla and Excuses crowd is going to assemble outside someones office tower. The someone will call the police, but the police will be busy, with the mess someones ECU peeping has made. The privat guards will do what private guards always do, if they see no good outcome for themselves and step aside.
The escape helicopter pilot will have lost loved ones in the mess and take off.
Someone is going to get shoved out the airlock, for all the other someones to see.
That will have a interesting learning effect.
Thus ended the lesson.
I'm increasingly convinced everyone is going to continue not to care, and then suddenly a 9/11-scale (or greater) event will occur, and it'll be 2002 all over again -- wild overreaction driven by hysteria.
Wide-spread internet-enabled command and control, plus long term surreptitious reconnaissance, coupled together with machine learning to maximize impact, and I'll hazard a guess that such a scenario can reach beyond the destruction of 4 full airplanes and 3 inhabited large buildings, in ways too uncomfortable to contemplate.
The impact of 9/11 was not just some airplanes and buildings - the impact was the whole of the Middle East in turmoil and a refugee wave not seen since WW2.
Not the main point of the article, but pretty interesting to see that Google's ad revenue is now apparently about equal to advertising revenue from all print media in the world combined.
I don't think we know where AI driven cars will lead.
no parking lots? If we have AI driven cars than billions of people who can't drive can suddenly use a car. Your 5yr old kid wants to visit grandma? The just ask the car to take them there. Grandma who's too old to drive wants to go somewhere? She jumps in an AI car. All that public transportation in Tokyo, HK, Singapore, NYC, London, Paris, Berlin? Maybe all that disappears. Maybe many people stay close to home because they don't have a car and public transportation is no fun (10 minute walk to station, 2-3 transfers, possibly a crowded train standing with sweating people, 10 minute walk from station to final dest) but if there are ubiquitous AI cars then I can go door to door at any time. Those two 10 minute walks disappear, the standing with sweaty people disappears. Now I start traveling far more often. Even better I don't have to concentrate on driving I can use that time.
I really doubt that. For many people, having a car is also a question of money. Also, in many cities it's more practical to use bicycles or public transport rather than cars. Also, the reasons 5-year-olds don't travel alone would still apply with AI cars: What happens if something goes wrong, e.g. the car breaks down?
You don't need to own an AI car to use an AI car. As for "what if something goes wrong" the same is true for any transportation. If I can stand anywhere in the city, press a button on my phone and have an AI car pick me up in 3-5 mins you can bet using them will become extremely common.
Public transport will continue to exist because it'd too expensive to build enough roads. London without Bus & Tube? You could walk faster than using a self driving car (recent tube strikes confirm that).
It would hurt medium distance flights and trains though. Boston-NYC would probably be more attractive in a self driving car than flying.
AI cars are supposed to solve traffic. That assumes all cars are AI cars but that's certainly one possible future (30-50yrs out). I'm sure someone thought horses would still be a thing when cars were introduced. I think it will go the same with AI cars. People will find the convenience too hard to resist and and some point I suspect human drivers will be relegated to race tracks and country roads
83 comments
[ 3.1 ms ] story [ 30.9 ms ] threadSelf driving cars on the other hand are right in front of our eyes, we have seen them driving around and are rapidly advancing in terms of their level of autonomy...in terms of timeline for widespread adoption, I'll look at utility. The utility of self-driving cars is just too great for them not to quickly replace manual cars once the technology is fully ready. I can't imagine too many people would want to be stuck fiddling with their cars in long commutes when they could be reading, watching TV or taking a nap...just as smart phones dominate now because of their utility, self-driving cars are likely to unfold the same way.
There's basically no evidence for this though. Sure Tesla can show me a 3 min video of a car driving in perfect weather on perfect, straight, wide roads but I don't see self-driving cars "in front of my eyes" at all. We don't have any safety data that is directly comparable to human driving, we have several accidents already, etc.
>when they could be reading, watching TV or taking a nap...
I agree that when/if we make it to that level of autonomy we'll have turned the corner on adoption and it will explode. But I'm not sure that even the most optimistic of self-driving companies would say that we are anywhere close to that.
Getting from "man powered craft that can stay in the air for 30 seconds and a hundred odd metres" to "jetting millions of people around the world every day" requires a leap of imagination, but unequivocally it was possible, and happened.
There is far more money in automating long-distance trucking. My bet is the first real use of self-driving "cars" will be for trucks. The human driver can sleep, eat a meal, etc but will remain on-board for safety's sake and dealing with unanticipated edge cases (e.g. the AI wakes up the driver when it runs into a traffic jam due to an accident. The AI can probably recognize cop cars, flares, etc but won't be able to navigate the scene of the accident while responding to instructions from police officers)
The next small step will be "follower" trucks. The lead truck has a human driver same as before, but one or two AI-driven trucks follow. They do their best to track the human driver's response to unexpected situations. Worst-case, they pull over and flag the human driver who pulls over and uses a VR rig to pilot them past the problem.
Meanwhile the application of "self-driving" technology to cars over the next 10 years will be to avoid accidents, not to actually be self-driving.
Will we have fully autonomous vehicles within 50 years? Almost certainly. Will we have them in the next 5? Almost certainly not.
http://www.riotinto.com/ourcommitment/spotlight-18130_18328....
I'd argue that the shit self driving cars we have right now are ALREADY statistically safer than humans.
You don't have to be perfect. Just better than an average driver. And average drivers are drunk, sleep deprived, and texting. That is a very low bar to pass.
Even if they were much more dangerous, there is still a huge market. Safely, is arguably pretty low on the list of things that consumers care about.
The problem with self driving cars right now is purely political.
But even that isn't insurmountable. All you have to do is release in a low government regulation red state, like Arizona.
And then the rest of the country will follow, once everyone in Arizona starts using them and realize how awesome they are.
I do think that AI and particularly deep learning are overhyped. It is a big deal, but it's not as big a deal as many people make it out to be. It's much like the Internet in 1998; it works, you can use it now, and you know that it's going to drastically increase the reach of computing power, but people are talking like that's all going to happen in 1-2 years when really it'll be 2-3 decades.
I guess when I look around, I don't see a comparison between AI now and the Internet then, I see a comparison between AI now and AI then (70's).
https://en.wikipedia.org/wiki/AI_effect
It's worth remembering that many of the actual technologies developed in 1970s AI - A*, beam search, pattern matching, constraint solving, rules engines - are now basic CS algorithms that most programmers are just expected to know, and they have continued applicability to a wide variety of domains. We just don't call them "AI" anymore, because the technology has been commoditized enough that they are "business rules", or "single-player games", or "layout engines", or other examples of the actual problem domain that's solved.
Ditto 1990s AI - fuzzy logic, SVMs, Bayesian networks, collaborative filtering, perceptrons. It just got good enough that it became the standard way you solve a certain class of problems with computers, and now it's called "fraud detection", "Zestimates (tm)", "recommendation engines", "stream ranking", and so on.
Deep learning is the same. The hype cycle will eventually bust, because all hype cycles eventually bust. But in the meantime, it has opened up whole classes of problems in computer vision, speech recognition, recommendation, machine translation, and several other domains that were not amenable to computerized treatment before.
That's how I see AI. People will surely get injures and even die in Accidents. We are a humanist society where human life is valued above everything else, so I'm very optimistic that AI will get better and give their humans more divine powers and longer life over time.
Well researched, fundamented, empirically valid views are worth pursuing, and if those happen to lead into contrarian, ok, nothing wrong with that, but pursuing it? No way.
GPS for civilians actually predates the 1996 directive http://www.pcworld.com/article/2000276/a-brief-history-of-gp...
Basically WebTaxi works well with people having wireless web access from PDAs (Personal Digital Assistants). Imagine a scenario where you are sitting in a restaurant and you need a Taxi service. You go to the WebTaxi website and request a Taxi to your home. WebTaxi comes back and tells you exactly how much it will cost you. It can even tell you how long you will have to wait for the taxi. You continue to chat while you wait for the taxi to show up. As soon as the taxi arrives, your PDA informs you that the taxi is waiting for you!
https://web.archive.org/web/20010127103500/http://www.eventh...
uhhh, wasn't I reading pieces about how the internet was really shaking up TV about 5 years ago?
The reason why I bring this up, as being important in 10 years or less. Is that it will be available for around $1m.
AI will negate the need to go through rigorous astronaut training. Maybe there will be some, but not as thorough.
Graphene and other exotic metals will allow for lighter craft and reusability.
High performance electric engines will allow for less fuel payloads to get into high altitude, whilst allowing for the metallic hydrogen to literally blast out into space. Metallic hydrogen is 7x more powerful than liquid hydrogen.
I think once average joe is able to get into space. We could finally see an abundance of rare earth materials. Could be like the old west, where a co-op of you and your buddies go out, wrangle an asteroid and mine. Come back with a trillion dollars worth of metals.
Of course, a glut of anything will bring prices down. But hopefully by then manufacturing is ubiquitous also.
Maybe it's the optimist in me, but there are certain historical events that could take our society from where it is now, scarcity based, to an abundant one.
Interesting times ahead.
https://en.wikipedia.org/wiki/Skylon_(spacecraft)
Personally I feel we are still more than 10 years away from mass applications. I'm happy to be proven wrong though.
Thinking from a Materials POV in the next 10 years I think 3D printing (additive manufacturing) is an area where large advances will occur. New applications for 3D printed materials and wider adoption of 3d printing.
I think that there's still ENORMOUS potential for mobile apps and mobile-driven services. There's huge opportunity for new services e.g. Uber. I think just about every large web/desktop property faces a significant threat from a mobile app e.g. craigslist, match.com (they luckily own tinder), flickr (wait, already gone to Instragam), AIM/Yahoo Messenger (gone thanks to Whatsapp), etc...
[1] https://www.recode.net/2016/9/16/12933780/average-app-downlo...
12 hours? Wow. TV maxed out at about 5 hours a day.
I'm actually surprised the 51% are downloading one or more a month, that's hundreds of millions of potential customers right there.
But they'll have legal pot to keep them quiet and passive.
[1] https://www.theatlantic.com/business/archive/2017/05/the-par...
No, they moved into factory, mining, retail, and other jobs that were commensurate with their abilities. The automation shift is not the same, because it will take over all the unskilled/low-skilled jobs, and there will be nowhere in the economy for the displaced people to move into.
It is not really conceivable that all these people are capable of taking on jobs which require complex thinking, judgment, creativity, or other higher skills. IQ is a real thing, and although it does not reflect on someone's human value, it does very definitely place limits on what functions a significant portion of the population can perform. And in the long run, we will all be in that displaced portion. Best that we find ways to create human security and meaning, now, for a coming world in which we have no necessary economic role to play.
There are plenty of "low-skill" jobs that nonetheless require human labor in some form because of the ability of humans -- even less intellectual humans -- to think creatively and contort themselves both mentally and physically.
Imagine a robot that does everything a school janitor can do. That simply won't exist in the near future. Sure, maybe the janitor won't have to mop large floors anymore. But there's plenty of other work to be done around a school nearly always to continue to justify their employment.
In terms of time needed to politically and socially prepare for that situation, 20 years feels really near term to me. I don't really expect our existing political structures and processes are capable of dealing with the challenge adequately in time. It's one of my top three fears for the <30 year future. And unfortunately the other two (global displacements from climate change, and bio-/nuclear-terrorism by desperate or angry people/groups) are synergistic with it. If it weren't for my normalcy bias, I'd be kind of terrified.
For how long, though? Is there any reason to think that will be anything besides a transitional stage before the robots entirely phase out the human?
I think a lot of people will be surprised at exactly which jobs too. It's not going to be the cleaners, the carpenters or the plumbers who still have to get into tight spaces that a robot butler can't handle. It's going to be the office workers that push computer bits around all day.
Any job where it all goes in and out over a wire is vulnerable. But robot manipulation in unstructured situations is still very hard, which means people are needed for their hands. I've been calling this future "Machines should think. People should work" for years. It's getting closer.
https://news.ycombinator.com/newsguidelines.html
Having said that, it's strange that this guy doesn't mention cryptocurrency anywhere in his post.
He spends a lot of time discussing technologies that aren't quite there yet, and leaves out the one area that is (likely) right at the point of adoption.
[0] http://a16z.com/2017/04/03/cryptocurrencies-protocols-appcoi...
Viewing habits are shifting but so is public opinion on Ads. The future of Ads on these internet TV sites is going to be the same as it was for traditional TV albeit a bit more targeted. Instead of TV networks controlling the viewing data in the future internet TV services will control it.
so the potential of a startup to take advantage of that data is going to be minimal since it will be guarded like gold.
(1) Software is well on the way to an old, slowly changing, mostly routine, boring, low margin, slow growth business and industry.
(2) That the general purpose computing hard/software and 100 Mbps Internet data rates are all so cheap will in places undercut large, expensive approaches to computing projects. E.g., people will do projects, but Sand Hill Road need not be more involved than they are now in, say, a new pizza shop in Peoria, not more involved than they were in the romantic matchmaking Web site Plenty of Fish.
(3) A huge problem for computing causing (1) is that there just are not enough good, new ideas for high value IT startups. For a joke, but a simple explanation, the days of a hot social, mobile, local sharing app are gone.
(4) For Evans's dreams about machine learning (ML), he is not very well informed and is not seeing very clearly, but in principle, if strain to extend the meaning of ML, he is correct about the potential. But in practice that future will come slowly, very slowly.
Why so slowly? In fact, e.g., looking at, say, scikit, that's basically some applied math, and nearly all the math there is now old enough to have long, gray whiskers, literally. That applied math and much more in optimization, simulation, experimental design and analysis of variance, stochastic processes, on and on, i.e., the math sciences, optimization, operations research, statistics, except for a few niches (e.g., optimization in operating oil refineries, the fast Fourier transform for oil prospecting, partial differential equations for modeling oil reservoirs, air line crew scheduling, maybe airline plane scheduling, statistics in biomedical research and development) hasn't caught on at all well generally in business; so there is only tiny reason to believe it will become "The next big thing" now.
For that math, it's not so easy to get good at it: (A) The pure math departments don't want to teach it. (B) Very few other departments have enough grounding in the math prerequisites to teach the material very well. (C) Except for a few niches, business has long made its position crystal clear -- such applied math and a dime won't cover a 10 cent cup of coffee. (D) It has long been the case and still is that about the only paying customer that takes such applied math seriously is the US DoD for purposes of US national security.
Run a good applied math project down Sand Hill Road, for a big problem, with running prototype software, and will hear laughs, silence, or "don't call us; we'll call you". I have to doubt if there is a single information technology (biotechnology can be different) VC firm on Sand Hill Road able to direct technical due diligence for an applied math project.
There can't be "The next big thing" without a few hundred times more well qualified startup founder, worker bee, applied mathematicians to do the work.
But also living in the city is going to get a lot easier because of robotic transportation. It won't be necessary to haul anything home from a store, for example. A lot of cabs could just have one or two seats and no trunk.
We'll see tiny vehicles delivering dinner, handling dry cleaning. And parking woes disappear.
But the biggest change will be in how easy it will be to move from one place to another. Yes, robots will be able to move all your stuff cheaply. But they also might just procure comparable stuff in your new city instead. And most of your stuff you don't use all the time, so maybe it's in long term remote/cheap storage until you do. Maybe its okay if those table leafs are at your house for Thanksgiving dinner.
Maybe you move every month or live in one place during the week and another on the weekends.
I think we should be less focused on autonomous driving, and more focused on reforming our tax code, infrastructure spend (too little), and healthcare spend (too much).
They haven't found it prohibitively expensive or unsustainable. It's worth it to them. Making that lifestyle cheaper and easier will expand the number of people for whom that is true.
Automation, as it has always done, will make our infrastructure cheaper and more robust. It will slash the cost of healthcare and dramatically increase the amount of resources we have available and our quality of life.
So, no, I don't think we should focus more on nibbling around the edges of what is possible. Those things can all be fixed, but that takes political will and convincing the majority of people. And most of them are stuck in the analogies of the past.
Student loans are driven by massive inefficiencies that various companies will exploit. The end will be apparent in 10 years.
The German "elite" label does not mean anything at all. It's basically a government program to allocate (a small amount of) additional funds decorated with a fancy term.
https://en.wikipedia.org/wiki/German_Universities_Excellence...
People can't really see it yet, partially due to a unwillingness to see the darker side of humanity, but a real cyber security expert is worth more than a cruise missile.
Planes, self-driving cars, swarms of aerial and underwater drones, fraud, propaganda, auto-targeting laser systems, surveillance, power generation and distribution, command and control interruption, machine learning + cyber warfare.
Who needs nuclear weapons when you have a zero day on a hundred million self-driving cars and you can get them to the nearest gas station or chemical plant going 300km/h?
To be fair I guess even with a non-connected car that I have to update via USB, I could still download a malicious update. But at least we won't all get the malicious update pushed at once.
People vote with their wallets, and if the technology-savvy and privacy-conscious HN crowd is voting this way then...
However, if someday a zeroday is used to duke it out, a very angry, very unreasonable, not interested PR-Bla and Excuses crowd is going to assemble outside someones office tower. The someone will call the police, but the police will be busy, with the mess someones ECU peeping has made. The privat guards will do what private guards always do, if they see no good outcome for themselves and step aside. The escape helicopter pilot will have lost loved ones in the mess and take off. Someone is going to get shoved out the airlock, for all the other someones to see. That will have a interesting learning effect. Thus ended the lesson.
no parking lots? If we have AI driven cars than billions of people who can't drive can suddenly use a car. Your 5yr old kid wants to visit grandma? The just ask the car to take them there. Grandma who's too old to drive wants to go somewhere? She jumps in an AI car. All that public transportation in Tokyo, HK, Singapore, NYC, London, Paris, Berlin? Maybe all that disappears. Maybe many people stay close to home because they don't have a car and public transportation is no fun (10 minute walk to station, 2-3 transfers, possibly a crowded train standing with sweating people, 10 minute walk from station to final dest) but if there are ubiquitous AI cars then I can go door to door at any time. Those two 10 minute walks disappear, the standing with sweaty people disappears. Now I start traveling far more often. Even better I don't have to concentrate on driving I can use that time.
In other words, AI cars might 10x car usage
It would hurt medium distance flights and trains though. Boston-NYC would probably be more attractive in a self driving car than flying.