AI in a post-capitalistic world

Having just finished Paul Mason's incredible PostCapitalism: A Guide to Our Future, I found it sparking in my mind many interesting ideas and thoughts on how artificial intelligence can fit into an economy increasingly dependent on the abundance of information. I must mention that I am not at all an economist, so apologies in advance if I confuse any aspects of the theory. Although Mason doesn't explicitly mention artificial intelligence, many of his ideas on info-tech are directly applicable and highly relevant, so I will first attempt to explain them.

Paul Mason envisions the postcapitalistic economy as a distributed network, without any central planner, but operating primarily on the free access and dissemination of information. The lifeblood of such a setup would be instantaneous, real-time information feedback, supported by a non-centralized technological network. His argument is that, as we shift increasingly toward an information-based economy, we move away from scarcity, because information is infinitely replicable. Something that is infinitely replicable has a zero marginal cost (meaning that after the first one is made, it costs pretty much nothing to make another). This makes it really difficult to determine its cost. 

In a free-market economy, prices are determined based on the scarcity of the item. For example, because there are only so many TVs available at any given time, if more people want a TV, the price will increase until supply meets demand. Supply can't increase immediately because there is a production cost for each additional TV. But with info-tech, this doesn’t apply. Netflix lets you share your account with multiple devices and family members without requiring you to pay more, because their additional cost per use (marginal cost) is practically nothing. The information itself, once created, can be freely copied as many times as needed to meet demand. Netflix is essentially charging for the fixed cost of creating the information aggregation technology (plus of course other things like advertisement, content creation, engineers, distribution, etc). Countless other examples exist of people/entities not even charging the fixed cost (i.e. Wikipedia, GitHub, open source) for access to information. 

So how do you price something like this? Paul Mason makes a good point that at present, the market doesn’t know how to price “information”. Hence the crazy valuations of startups that aim to aggregate and deliver services based on it. The most we can do is compare with prices of services without info-tech. For example, if I book a hotel room, it will cost me X dollars, but if I can find someone willing to let me stay in their home while they’re gone for Y dollars, then the value of the service that allows me to find that someone should be close to X - Y. But what about the price of having songs recommended to me on Pandora or Spotify? I would not have paid to listen to these songs otherwise, as I didn’t even know they existed. But surely this service is worth something to me. What about something I don’t even have to explicitly pay for, such as Twitter or Reddit? What is the value to me of having in my hands a snapshot of current trends and articles, tailored to my specific interests? 

With capitalism's inability to appropriately price information-based goods, along with the ease of copying information, Mason argues that the cost of labor will be driven to zero, thereby producing enormous profit for corporations. This is feasible until almost all the labor has been squeezed out of the system, at which point profit flatlines and prices are driven toward zero because nobody can afford the products anymore. (This is my very layperson understanding of labor theory.) Therefore, in a post-scarcity world, capitalism cannot exist.

I found this a fascinating idea, but I have a major issue with this contention. Information cannot reduce the cost of resources to zero. In fact, nothing can. We still need to make physical products such as housing, food, clothing, etc. Info-tech can perhaps make resource consumption much more efficient, such as encouraging telecommuting or reducing overproduction, but it can never drive the cost to zero, in the absence of infinite resources. 

However, his point is well-taken, because the second thought I had was that the replacement of labor is primarily going to be by artificial intelligence. Mason doesn't really mention AI in his book, instead simply referring to "automation," but he seems to assume that once a process automated, it essentially becomes capital. It cannot ever inject more value into the system. But AI appears to be making remarkable strides, and is predicted by many to cause massive unemployment in the near future (especially for routine work, see below, and in the automotive sector). As artificial intelligence replaces labor, most people will be driven out of the market. 

This will be a problem (to say the least), because postcapitalism assumes that we as individuals can retain control of social and other information. But most companies are extremely protective of their information (as they are essentially smart information aggregators). You can use Uber to get a ride, but you cannot have access to their database. (It is this information imbalance that gets converted to profit.) So with corporations owning both the aggregated information as well as the algorithms to process that information, the replacement of human labor could lead to an economic disaster. Perhaps, in the face of such widespread unemployment, social pressures will be so large that governments will be forced to institute a basic universal income. This is vigorously advocated by Mason as being one of the central requisites of a postcapitalism economy, but it seems like there will need to be incredible social upheaval before this will be considered. 

Another interesting idea that the book brings up is that, in an information-abundant, technologically connected society, profit derives primarily from creating something. In this world, anything that can be planned can be automated. Therefore, the only jobs left will be those that are inherently innovative, creative, and unplanned. But we are already starting to see AI that can create art, generate Shakespeare, and even dream. True, there is still a long way to go before we recognize such output as truly innovative or useful to daily life, but we are only beginning to explore the full potential of deep learning and other relatively new AI methods. Will it just be a matter of time before humans are competing with machines to innovate and create? And if/when that day comes, will it be ethical for anyone or anything to "own" these creative artificial intelligences?

Overall, I found PostCapitalism to be a great read. Highly recommended for anyone looking to get a better sense of how our economy is being transformed. Mason doesn't go into detail about how exactly we are supposed to make the transition to postcapitalism, but he makes a great case that certainly something interesting is on the horizon.

Apple vs the World: On the broken state of patent law

Image from Cult of Mac

Image from Cult of Mac

(originally posted at amaral-lab.org)

On August 24, 2012, Apple won a huge patent lawsuit against Samsung and was subsequently awarded $1.05 billion in damages, an amount that they are now seeking to triple because the infringement was found to be “willful.” However, this is only the latest battle of a huge litigation war waged by the two tech giants in multiple countries around the world. It also sets the stage for a larger attack on Android, which Steve Jobs in his autobiography claimed was stolen and that he would try to destroy (although I honestly don’t see how anyone could mistake iOS for Android).

The patents that were found to be violated: pinch to zoom, bounce back from scrolling, tap to enlarge, multi-touch sensing, rounded square icons, and the general shape and look of the phone.

The implications from this verdict could be far-reaching. In this modern-day technological age, IP (intellectual property) is becomingly increasingly hard to define, especially given the current state of patent law in the US, which many have decried as being broken. How do we determine the validity of a patent for “swipe to unlock” when it’s the digital equivalent of “turn doorknob to open door”? Incidentally, Apple lost a suit against HTC in the UK over this particular patent, which the judge declared as “obvious.”

Does this mean that tech companies shouldn’t patent? It’s telling that most IP scuffles in the tech industry end in settlement, with both sides agreeing to share a bit of their intellectual property in a fair manner. This is because the industry is so interdependent, and patents exist for everything. Any one device uses scores of patented technologies and hardware that are all essential to its function, and no tech company could possibly exist in isolation. I for one find it highly amusing that Apple itself is Samsung’s largest customer, purchasing over $5 billion (estimated to be as high as $11 billion for 2012) in processors, screens, storage, and other parts from it every year, and Samsung is Apple’s key supplier. The two literally could not survive without each other. Apple’s own iOS uses several technologies that Samsung invented, a fact which Samsung tried to use to countersue, although these efforts proved ultimately unsuccessful.

Let’s just say the Facebook relationship status of these two would be “it’s complicated.”

What is the fall-out from this trial? Most think it’ll lead to many more legal battles, since a precedent has been set that such battles can be won and are highly profitable. Phone makers will shy away from the current standard of smartphone designs, which Apple did indeed pioneer and establish. This could drive innovation, but could also create unnecessary obstacles if phone makers feel they can’t use common-sense features such as “rounded corners” or “slide to unlock.” There could also be a gradual shift away from Android OS, which is the ultimate target of the war Apple is waging. Google realizes that it can no longer afford to sit on the sidelines. The company has already stepped into the fray by capitalizing on its recent purchase of Motorola Mobility, filing a lawsuit against Apple for infringing on a wide range of mobility patents it acquired through the purchase.

Essentially, in an industry 100% dependent on constant innovation, it’s hard to define what is truly innovative, and how much of that innovation rests on what came before it, which is basically a credit assignment problem. How do you determine what is “obvious,” which is to say that this idea would have been independently reached by most people because it naturally follows from what already existed? On the flip side, how do you determine what is “innovative,” which is a significant, novel contribution that would not have existed otherwise? And is it fair to expect overworked, time-limited patent employees to answer these tough questions, answers that are literally responsible for the outcomes of billion-dollar lawsuits around the world?

I don’t have answers to these questions, but I do believe that the patenting of cosmetic features is completely frivolous and wastes both time and money. It’s interesting to compare the tech industry to the fashion sector, which is also predicated on never-ending reinvention and innovation. What would happen if we started to patent “green dresses with zig-zagging dark patterns” or “high-waisted plaid shorts worn with belts”? If it’s ridiculous to patent these design concepts, it should also be ridiculous to patent the “rounded rectangular shape” of the iPad or the “round square icons on a black background” on iPhones (patents which both exist). In fashion, just as in technology, you pay a premium for the brand. Therefore, it’s illegal for knock-off Gucci products to claim to be actual Gucci, but it’s perfectly fine to create discounted items which look highly similar, but don’t carry the same brand name. If Samsung had slapped an Apple logo on all their products and tried to capitalize on their premium brand name, then this would be a clear case of stealing. But technology, just like fashion, follows trends, and the current trend in smartphone technology is to have large touchscreen rectangular phones that are application-based. It is my opinion that you can’t patent trends. And chasing after perceived patent infringements could very well come back to bite Apple in the ass. Because if there’s one thing we know about fads, it’s that they all eventually die away.