Google Will Become an AI Company

For quite some time I’ve been pretty down on GOOG. Not the company, Google, but the stock in it. The reason, I realized, is that I’ve been thinking of Google as a search company and search as a market is largely played out. There’ll be some growth from mobile, though much of it cannibalistic, and some further increases from local and just general demographic shifts as the web insinuates itself further and further into more people’s lives. But overall search will remain an industry measured in the tens of billions for quite some time, and their stratospheric stock prices have left better investments in many other corners of the market.

My thinking might have been wrong though because Google isn’t a search company. Right now they’re a product development company (and generally a poor one) funded by search. In the future, though, they’re going to transition to an artificial intelligence company. They’re already halfway there. Search is just the problem they first applied AI to, and the one they’ve grown to dominate. In the future they will have much bigger fish to fry. While Facebook is still trying to figure how to make money off of people sharing pictures, Google may be revolutionizing one industry after another by pushing the bounds of what software can do.

Take the automotive industry. That’s nearly two orders of magnitude larger than search. To put that another way, if Google managed to scoop up just 2% of that industry they’d have more than doubled their revenue. With their driverless car project, I think they’ve got a shot of taking a much bigger slice of the pie than that.

Imagine a world in which all cars drove themselves. Seriously think about the ramifications. Here are just a few:

1. Cars may be cheaper and/or higher markup. Mechanical drivers will eventually reach a point where errors are no longer a common occurrence, and as a result, safety regulations (which currently add significant expense to cars) could be greatly relaxed. The conversion to electric cars (which will have to occur, as fossil fuels won’t be able to accommodate the meteoric rise in drive time) will also eliminate expensive exhaust systems.

2. Children could own cars. Don’t feel like schlepping your kid to soccer practice? Just buy them a car. Age restrictions on driving only exist because children can’t be trusted not to kill themselves or others on the road. If a machine is driving for them that’s no longer an issue. Parental controls will easily alleviate other concerns.

3. The beverage industry will grow. Designated drivers are a thing of the past. Go to the bar, get wasted, have your car take you home. Hell, get a rum and Coke to go on your way out the door. How many times have you had a beer or two less than you wanted to because you knew you had to drive home? Never again.

4. Speed limits will be unnecessary. A mechanical driver processes information so close to instantaneously that mechanical limits will be the only factor restricting speed. Cars that can travel at 200 mph will become common and fetch premiums.

5. Traffic, too, will become a thing of the past. Slowdowns are caused entirely by human error. Accidents, people stopping abruptly, merging poorly, etc. Intelligent routing and better driving will mean that you’ll maintain top speed throughout your trips.

6. The map will shrink greatly. Right now I live about 30 miles from my office and the commute is on the very edge of what I can stand. Make my car driverless (freeing me up to watch TV, read a book, catch up on emails, etc.) and able to travel at twice the speed, and spend the entire trip at top speed (rather than slowing down and speeding up on the highway) and I could feasibly live as far as 100 mph away. Since the area of a circle is proportionate to the square of the radius my possible housing locations just grew by about 11x.

7. Urbanization will reverse. Why pay $3,000/month for a flat in Manhattan when you can get from 100 miles upstate to work in 30 minutes? That’s enough time to watch The Daily Show on the way in anyway.

8. Airlines will be devastated. Why fly from New York to Chicago? Just hop in the car, watch a couple movies (on the screen that is mounted where Wikipedia says that something called a “steering wheel” used to be) and you’re there.

9. Other forms of public transport won’t fare much better. A driverless cab won’t cost much more than a bus (which also will be driverless) but will be a hell of a lot nicer. Don’t even get me started on subways.

I could go on but you get the point. Whoever invents the driverless car is going to make a lot of money. Possibly more than anyone has ever made before by an order of magnitude. This could be Google. I’m not saying it will be them. There’s far too much standing in between us and the driverless future to predict something like that accurately, especially since many of the obstacles are governmental. But it does seem as if it almost has to happen eventually, and right now they’re probably the frontrunner.

And that’s just one industry. Imagine customer service. A program that could do as good of a job as a real call center rep could simultaneously decrease costs and improve service. It would replace millions of employees nearly overnight. How much could that make? There’s almost no industry that can’t be greatly improved by AI.

So while as a search or mediocre product development company GOOG is starting to like more like a blue chip, I think as an AI company, Google has a shot of generating massive growth.

About these ads

51 Responses to “Google Will Become an AI Company”

  1. Wouldn’t an AI controlled aircraft allow for cheaper and safer air flight? In that case shorter range and smaller capacity air travel may become more reasonable and thus air travel may not become as antiquated as you suggest.

    • Sure. AI controls much of the flight already (not to mention the explosion in popularity of UAVs) so it’s easily conceivable. I don’t think too much of the expense of a flight comes from the pilots though, especially on the regional carriers like Southwest and AirTran. Hardware, maintenance, airports, fuel, etc. should keep air travel pretty pricey.

    • breathethereader Says:

      At the risk of upsetting the apple cart here, I thing most of the comments here reflect exactly what companies like Google want to see. All of this talk about advanced technological changes that might further the automation of machines we build, like cars for example, reflects a level of social unawareness and complacency that is already allowing Google and its “social networking” friends to turn us into the automatons – a population of mindless sheep in a quietly complacent stupor. We are willingly permiting the largest “surveillance network” ever conceived to strip away our personal freedom and privacy…in fact, we are allowing this humongous Peeping Tom to strip away our memory of those personal freedoms.

      When Public Assistance recipients collect I-Pods, Androids and truckloads of other devices paid for by taxpayers and allow them access to the articial world known as the Internet, we know that the priorities once inherent to America’s extend a helping hand to the “have nots” have changed. Right now, as I write this, millions of so called underpiveleged people are trading off food stamps for cash – money used to empty the shelves at Best Buy of chip driven toys.

      Welcome to the Matrix, everyone…and batteries are included.

      MK Dion
      check out “Mice” at Breathe The Reader

    • breathethereader Says:

      At the risk of upsetting the apple cart here, I think most of the comments here reflect exactly what companies like Google want to see. All of this talk about advanced technological changes that might further the automation of machines we build, like cars for example, reflects a level of social unawareness and complacency that is already allowing Google and its “social networking” friends to turn us into automatons – a population of mindless sheep in a quietly complacent stupor. We are willingly permiting the largest “surveillance network” ever conceived to strip away our personal freedom and privacy…in fact, we are allowing this humongous Peeping Tom to strip away our memory of those personal freedoms.

      When Public Assistance recipients collect I-Pods, Androids and truckloads of other devices paid for by taxpayers and they now have access to the articial world known as the Internet, we know that the priorities once inherent to America’s commitment to extend a helping hand to the “have nots” has changed. Right now, as I write this, millions of so called underpiveleged people are trading off food stamps for cash – money used to empty the shelves at stores evertwhere of chip driven toys. So what if their children don’t eat?

      Welcome to the Matrix, everyone…and batteries are included.

      MK Dion
      check out “Mice” at Breathe The Reader

  2. I don’t know if anyone would actually own a self-driving car — it makes more sense for you to pay for one as you need, or subscribe to a car service for X dollars per month. The way our cars now sit parked for 22 hours in a day is a waste. Instead, I’d tell Google Cars that I want to be picked up in 15 minutes from 123 Main St, and it’ll route the car that will get me there the fastest. Carpooling would save even more resources. For a 25% discount, I’d be sent a car that includes a few other people going to nearby places. Adding only a few minutes to my trip, I would deem the cost in time worth the money I save.

    • That would make a lot of sense. I could see both really, car ownership as a luxury, rental as the economic choice.

      I’d personally still prefer to have one, it’s sort of like a rolling storage locker. I might not pay $50k for that privilege.

      • Yes, all these rolling storage lockers carelessly cluttered around my neighborhood really get me down :-) whose idiotic idea was that, anyway.

        you’re not allowed to leave your crap laying around outside in front of someone else’s house unless it has 4 wheels under it and is reasonably automobile shaped… why is that, anyway?

  3. I see one huge stumbling block there: good old fashioned fear-mongering, funded by companies who stand to loose. Once driverless cars become a reality people will say they need to be outlawed and use specific cases of failure (and these driverless cars will fail occasionally, as all technology and humans do) as examples.

    The rational way to do this is to have a neutral regulation: establish how safe the average human driver is, then require any auto-pilot car to be safer. Unfortunately, facts and reality have little to do with politics and policy, so it’s probable that you won’t see driverless cars in your lifetime.

    • There might be enough companies who would stand to profit that the lobbying will come out in favor. Certainly whoever invents the cars. Possibly the auto industry (depending on what they think the effect on sales will be). Most definitely big oil, or whatever replaces oil (which will have to be replaced with a substantial increase in mileage.)

      • theorange Says:

        As someone who actually DID loose the ability to earn money due to the sub-prime lending practices of the establishment and the corrupt politicians sitting in office w/o term limits, who by the way happen to be 97% lawyers, I look forward to anything and everything that can put an end to their form of making a living. Period. If we didn’t have scum-suckin, ambulance-chasing lawyers, most ALL of our problems would be evaporated. Imagine if you were an auto mechanic.. Would you look forward to a car that didnt need repairs? No. You wouldn’t. And neither do our politicians who NEVER get rid of laws, but only add to them. Legalize Cannabis is one simple example. Rant off.

    • Billy Wenge-Murphy Says:

      But they won’t be invented in one fell swoop like happens in sci-fi. We’re already in the process of inventing the driverless car, right now. It’s being invented incrementally every day, and part of it is already on the market. There will be no single turning point where it suddenly appears, so no company will be able to jump up at that dramatic moment and fight it. Rather, companies are competing TO bring us the pieces of this invention, because they can profit at each step.

      Alternative energy is a good place to look for a concrete analogy. Oil companies are fighting against an all-at-once sea change (which we could enact today if they weren’t in the way) but at the same time, they’re helping develop this technology because they know it’s inevitable and they want to gobble up the patents (as well as improve their image by building good will. Supporting green tech = warm fuzzy feelings = $$$$$)

  4. Kit Plummer Says:

    Pilot-less cargo aircraft are first. Then pilot-less commercial aircraft. We’ll likely not be seeing autonomous ground-based vehicles until after then. The FAA will be the guinea pig for all autonomy legislation required before the technology will have a chance to leave the lab.

  5. Think about parking, too: Certainly a lot of purchasing has moved online, but we’ve all seen the mall around the holidays. Post-AI driving, who cares? My car can drop me off at the door and go park itself. I can open up my phone after I’m done shopping, and it will be waiting for me by the time I get outside.

    And if you love the City, but hate parking: no worries. Your car knows you set your alarm for 6 AM and have an appointment at 8:30. Even if it parked an hour away, it will be by your door when you’re ready to go. Even if it’s right outside, it will be warm and ready when you need it.

    Does your wife go in at 9, while you start at 7? Right now that’s two cars. Post GCar, you can get your ride in and then send the car back to pick up the wife. Most 2-car families I know have two cars for these sorts of reasons. Will that be necessary post-smart car?

  6. Anonymous Says:

    1. Decide you want to kill yourself
    2. Jump over fence and on to highway
    3. Get killed by car traveling 200mph.
    4. Fuck up the whole beautiful system

    • So, you reckon that an automated AI with some kind of image recognition algorithm that recognises shapes of people walking into the middle of the road (necessary for driving in town anyway) and a cold, calculated reaction would do worse than a drowsy human driver’s emotional reaction in the same circumstances?

      On a road filled with auto-cars, I suspect if you jump into the motorway the cars will just shift lanes metres before your predicted location to make sure that they don’t hit you, without even slowing down – unlike the human drivers, who would probably slam the brakes and cause a pile-up. Killing yourself this way might be quite tricky.

      • Jouni Osmala Says:

        So people can play real life frogger with autocars. Run accross the roads safely in groups and and do all kinds of crazy stuff that today won’t happen because they are affraid of human error.

    • Not true. Unfortunately, people commit suicide on the DC Metro. Fortunately, this has not destroyed the public transit system in the District.

  7. Actually driverless subway trains has been in operation for a while already, here’s an article with photos from a chinese driverless subway line: http://english.cntv.cn/20101022/101856.shtml

  8. [...] Google Will Become an AI Company, mattmaroon.com [...]

  9. Google already is an AI company. “Goog” is the host of http://code.google.com/p/mindforth/wiki/AiHasBeenSolved and many other open-source AI projects.

    • Google isn’t an AI company yet, they’re a product development company that often uses AI. Calling them an AI company is like calling Procter and Gamble a packaging company. It isn’t their main focus, it’s something they do to achieve goals.

      My contention is that Google’s focus will shift entirely toward applying AI to problems.

      • GOOG _is_ an AI company at its core. Relevant search results were followed by relevant ads, voice recognition, and machine translation translation, all AI at some level, It’s not like calling Procter and Gamble a packaging company — it’s like calling P&G a soap company: soap is at the core of many of their products.

  10. Facebook is already an AI company.

  11. There’s already a company building AI-powered virtual call center reps – SmartAction (http://www.smartaction.com)

  12. The Google leadership has repeatedly said that search is an AGI hard problem. The social graph is also an AGI hard problem, for what it’s worth.

  13. LOL. You forgot the flying cars in every garage.

    There are far larger obstacles than you think. AI isn’t magic and doesn’t suspend the laws of physics or queuing theory (or human nature). It’s going to be a while before you car recognizes pedestrian body language enough to navigate around pedestrians in the city, much less thinking it can do the same at 200mph! Extending your drive range to 100 mi means huge expansion of infrastructure, because even an AI driver can only get so many cars/minute/lane. You can’t safely mix 200 mph cars with others unless your AI can also suspend inertia; the AI driver may know that it’s about to crash but still be unable to keep it from happening. Sending the same car back and forth may save on metals and plastics, but it means doubling the traffic with half the trips empty. You expect electric cars to be much cheaper because they don’t need an expensive exhaust system? Read more about the respective bill of materials, you have your figures wrong. Fewer safety provisions? More likely they will need more. Have airplanes gotten cheaper and needed less safety as the autopilots grew smarter?

    And of course, our current simpler technology works with very few problems, right? As we move more towards complex interactive software systems, bugs and malware have pretty much dissappeared, right? How long since the last time you had a “crash” or temporary freezup? A network delay or error (remember that all the cars need to cooperate, along with signals and sensors).

    There is also a question of responsibility. Suppose that a given AI driver is claimed to be somewhat “safer than the average driver”. If a given average driver does cause an accident, there is somebody to be held accountable. Who is responsible if the autodriven car has an accident and harms somebody? The passengers or vehicle owner? The mechanics? The manufacturer? The network operator? The street maintenance folks? It’s going to take a while to sort this out, and a very slow incremental learning period to separate the hype and “lab tests” from real world empiracal data.

    I could be wrong, Feel free to invest your life saving in the expectations you have for flying cars of the future, soon. I believe that AI driven cars will take many years to become common, and will even then occupy certain limited (but perhaps interesting) niches where they go on automatic only in special lanes of a few highways with only other automated cars.

  14. [...] Google Will Become an AI Company For quite some time I’ve been pretty down on GOOG. Not the company, Google, but the stock in it. The reason, I [...] [...]

  15. Everything they do is about Big Data. Collecting as much information as possible, and then throwing a lot of algorithms at it to pick out interesting points.

    For example, they launched a free 411 service called GOOG411 – why? To provide a simple way to crowdsource a huge volume of samples of people’s voices and questions. This eventually went into powering Google’s mobile search.

    Google isn’t going to become an AI company – it already is.

  16. Two major problems:
    1. Transition. 2. Security.

    Both can be handled by one solution. Obviously you cannot have any road with both human driven cars and self-driving cars. So transition is an issue.

    And if cars are self-driven, then the possibility of sabotage of the road becomes a question–especially at high speed.

    I’m thinking we’ll begin building underground highways for these self-driving cars.

    And when the above-ground freeways are empty of cars they can be used to harvest sunlight.

    Lastly, side-roads will still exist. Just, once you get close to your destination, you’ll leave the underground fast-track and take the slower-speed rated side-roads that already exist to your destination.

  17. [...] Google Will Become an AI Company « MattMaroon.com [...]

  18. [...] Google Will Become an AI Company « MattMaroon.c&#111&#109&#10LG KF240 Mobile Phone Features & Price in India [...]

  19. Jojomonkey Says:

    Fun and interesting ideas. But I can imagine that happening more in emerging countries that are actually interested in changing their infrastructures and are not caught in so much red tape. I can see China or Japan doing that faster than we do. When was the last time America did something radical and for the better? I say this because if GOOG wanted to do the above it would need serious govt support to shield it from the vast army of other companies that would expect to lose market share from such infrastructure changes. Do we even have 1 bullet train in the states?

  20. [...] Google Will Become an AI Company « MattMaroon.com [...]

  21. inboulder Says:

    This would have been a interesting article if written in the late 1860s. I believe thousands of sci-fi writers have covered the various implications of driver-less personal transport, does your article add anything to the discussion?

  22. [...] Google Will Become an AI Company « MattMaroon.com [...]

  23. breathethereader Says:

    You may be correct, but I have a different perspective that you might wish to consider. While it’s certainly possible that Google may become an AI company, (check out my blog entitled “Mice”) I’ll keep this short by simply telling you what Google and the enire “social network actually is right now – the world’s largest and most powerful survelleilance network. And as a whole, America’s blogging/messaging population is blindly allowing their freedoms and personal privacy to slip away, in their semi-stuporous acceptance of the obvious.

    For what it’s worth, however, I find your blog interesting and informative.

    MK Dion
    BreatheTheReader
    http://breathethereader.wordpress.com

  24. Thinking of driverless cars, I think Google is trying to invade into areas where it doesnt have ‘core competency’. I would have expected this driverless thing coming from a consortium of car manufacturers -imagine every new car being built with electronics allowing it to navigate by itself amid other cars with likewise chips as a unified, disciplined group that only needs to obey traffic rules while consistently maintaining optimum routing for all. Once a car weaves in to another group of cars on the move, it needs to adjust to all the others’ current and expected position (route and trajectory), speed and direction, then find and keep a position as ‘one among many’, adjust speed and acceleration to maintain constant velocity in given direction, etc. so every car still maintains optimum movement toward their respective destinations. We need to develop ‘smart swarm robotics’ algorithms with specific reference to self-directing traffic if these algorithms dont exist already, where members could continually join and leave a group, number of members is unknown, and other members need to constantly adjust and adapt their operating parameters in wake of the effects such a setup can produce.

    As regards AI use in search engine, I think Google are there already. Google seems to be able to ‘sense’ what you are about to type and 95% of the time its the same thing you had in mind. I can only think of including AI in the mix as a way to carry on conversations with a search engine so it can progressively refine the search result set towards returning only the most relevant ones. I can only think of being able to specify ‘what I dont want’ rather than the current way of ‘what I want’. Some of this filtering is currently being done by the options in the left hand menu. Filtering and trimming is the only known way to return the most relevant result possible. What are the modalities and how will this work out? I have no idea. I will leave that to the PHDs that Google employs.

  25. [...] Google Will Become an AI Company [...]

  26. Interesting observations, but this still leaves the largely untapped market of Algorithms for Artificial Stupidity (AS). Given that by definition half of the population live below the mean, we need robust research into AS to replace all occupations. For example, these cars need to occasionally run red lights so the artificial policebots don’t get too rusted up sitting around waiting for action.

    Only when the full gamut of human professions are replaced will be able to relax and eat potato chips as our driverless cars take us between Hard Rock cafe and Disneyland franchises.

  27. catholiconahottinroof Says:

    Interesting.

  28. Matt Murray Says:

    What about those craaaazy people that enjoy driving? I for one really like driving to and from work. I would certainly not be an early adopter of this tech. So if there are AI cars on the road with us normal folk, then there will still be the traffic jams etc… this world of the perfect commute lives in Minority Report and the Jetsons.

  29. [...] has been much written about Google’s spam problem and that Google is becoming an AI company. I just realized recently that search is AI, it is actually [...]

  30. [...] Google Will Become an AI Company – “For quite some time I’ve been pretty down on GOOG. Not the company, Google, but the stock in it. The reason, I realized, is that I’ve been thinking of Google as a search company and search as a market is largely played out. There’ll be some growth from mobile, though much of it cannibalistic, and some further increases from local and just general demographic shifts as the web insinuates itself further and further into more people’s lives. But overall search will remain an industry measured in the tens of billions for quite some time, and their stratospheric stock prices have left better investments in many other corners of the market….” [...]

  31. [...] Google Will Become an AI Company « MattMaroon.com [...]

  32. What are the most ‘future friendly’ web application frameworks to select in 2011?…

    Mobile Javascript is the platform for 2011, even if you are considering rich internet web applications. Here’s why. 2011 is the Year of the Mobile Apps [1]. This is highly significant because the iPhone represents the first major shift in computing fo…

  33. Where will you put all these cars for all these new drivers? Sounds like a traffic nightmare. More wars to secure cheap gasoline? Public transit must be funded properly and increased. Subways, high-speed rail and more commuter lines in more areas is the best, most efficient method—if they are funded and the tax structure is friendly toward it.

  34. [...] observations on the future of Google as an AI company. But it leaves open the largely untapped market of Algorithms for Artificial Stupidity (AS). Given [...]

  35. When a driverless car is invented the economy will collaps. No trafficlights, no signs, no sideprotection, no police, no parkinggarages in town, no parking guards, no busses, no trains, no new carmodels, no different engines, etcetera. I suggested such a plan in the Netherlands 15 years ago. It is such a big change that too many people will resist such plans and kill them. Dreaming is fine but even Google will not succeed. I hope that Google will work on changing the minds of people into a cleaner and better future.
    ceesdevrieze, Bloemendaal, the Netherlands

  36. Why Google? I would suggest a company such as IBM which has a lot more experience in industrial automation, much more research in the relevant area, and some AI experience.

    Anyhow, this future is still quite far off. You call Google a search company (which I would suggest they aren’t, they are an ad company first and foremost) but they diversified that long ago. Today they are big in outsourced Internet services with Google Apps, and they will drive this with Android to market dominance.

    They are also big with cloud hosting with App Engine as well and will quite possibly be one of the bigger players here as well (together with Amazon, Rackspace and a few others).

    That’s two more immediate markets which will grow considerably within the next decade and where Google has at least a chance of chance of dominating.

  37. [...] Google will become an AI company. Will it rename itself SkyNet? Seriously though, I think they’ll need acquisitions to make that happen, and really they need to start with a major overhaul or fix to their search algorithms as they are flooded with content spammers. Duck Duck Go might be interesting, though their value proposition is not following your every move, so I think they believe Google already is an AI company. [...]

Comments are closed.

Follow

Get every new post delivered to your Inbox.

Join 31 other followers

%d bloggers like this: