What I learned from building a neural network. Hint: the robots are coming!

I’m taking a developer certification for using IBM’s Watson AI, and one of the learning requirements is to understand the basics of artificial neural networks. In order to retain the information better and to understand the underlying processes, I decided to actually create a neural network, with the help of Stephen Welch’s excellent “Neural Networks Demystified” video series. You can see part one below:

I honestly did not expect it to be so complicated. Of course, it’s machine learning, it’s not supposed to be easy; but still, the amount of equations that described even the basics of a neural network were…out of my comfort zone to say the least. Nevertheless, it was eye- opening. Artificial neural networks (ANN) are a mathematical and programmatic representation of how neurons and axioms work. I am not going to delve into the mechanics of it, but it suffices to say that these ANNs are the beginnings of a general artificial intelligence: one that can think, understand and display intuition.


A demonstration of how Artificial Neural Networks mimic real, biological neurons. Source: InTech

The implications for this kind of technology are profound. It got me thinking about the economics of implementing such a system, only to realize that we are already in the midsts of a global upheaval thanks to the introduction of machine learning algorithms.

In 2011, Marc Adreessen, an early investor in Facebook, Twitter, Pinterest, and many other Silicon Valley “unicorns,” wrote:

“Software is eating the world.”

His statement still holds true, but I’d change it slightly to say, “AI is eating the world.

Unfortunately, the general public’s conception of AI is limited to Hollywood movies, and is almost completely abstracted from the real-life implementations of this technology. Many are unaware of how much this technology has infiltrated their lives. You can attribute your Netflix binging and endless Youtube video watching to the power of machine learning algorithms providing you with “suggestions” and “recommendations.” These services profile every move of yours, every bit of information, to pinpoint your demographic and provide you content that statistically fits with other people like you.

Yes, in AI, you are just a statistic.

cs humor

But AI does much more than that. Look no further than autonomous cars, self-running factories in China, and virtual assistants to see how this technology will seep into every industry of the market.

With such a powerful tool in our hands (quite literally), it is unfortunate that the labor market, and the institutions that feed into it, are not prepared for this transformational change. Most universities don’t have AI programs in place. Coding is still seen as being in the realm of engineers and nerds. Companies still operate with old OS versions of Microsoft Vista and use fax machines to exchange information. A large portion of the economy is simply lagging behind when it comes to it’s ability to change and adapt to an AI-based economy.

Now, this is not all fault of their own. Artificial Intelligence is a very complex subject, as I initially discovered. It requires advanced mathematics, advanced programming experience and a good amount of years in the practice to develop an effective AI architect. The amount of resources invested to produce such a focused individual is akin to the training regiment of a special forces soldier. It takes a lot of time, energy and talent to produce this worker of the future. However, such a worker will become indispensable for the future economy.

AI is like having a self-replicating mind. Another mind that does not need to be fed, does not sleep, does not complain, does not need health insurance, and is millions of times more powerful in mathematical computation than any person alive. It is the virtue of a capitalist society to employ such a tool if it deems it economically advantageous. It would be illogical not to employ it.

But herein the crux of the matter: A few amount of people will be extremely productive in the creation of wealth thanks to their use of AI, but what will become of everyone else?


Greater productivity is the holy grail of economics. It means the country can produce more, for less, at a faster pace. Global productivity exploded after the industrial revolution, thanks to industrial machines. Then, it sharply increased again with the advent of computational machines. Now, it’s due for another increase with the advent of commercial artificial intelligence. Here are three reasons why I believe the rise of AI is bad news for the global labor market:

1) Job replacement will happen faster than job creation

2) Productivity will be focused in a corporate oligopoly

3) “Enormous Data” will provide these companies a competitive advantage over the rest of the market



Job Replacement

This is a big one, especially since it’s become so politicized in the last couple of months. Jobs are always replaced by the coming of newer technologies. When the car became mass produced, the horse carriage industry (the traditional mode of transportation for centuries) underwent an irreversible decline. However, the collapse of this industry was supplanted by an even greater upswell of economic wealth created by the car: stables were supplanted by gas stations, horse drivers by valets, streets needed to be paved, cars needed to be maintained, manufacturing increased in order to keep up with the demand, etc. Thus, older technologies are usually supplanted by newer ones thanks to the new jobs it creates.

In an article for MIT Technology Review, Joel Mokyr, a leading economic historian at Northwestern University commented on the increasingly fast pace of disruption:

The current disruptions are faster and more intensive…It is nothing like what we have seen in the past, and the issue is whether the system can adapt as it did in the past.

He further states how jobs that require automation -usually reserved for the lower classes of workers- will be the most susceptible to this change. If these workers are to keep their jobs and adapt to the new AI economy, they must obtain a degree in computer science or a similarly technical field, as well as a specialization in whatever field they will be working in. This kind of education is expensive, and it falls within the responsibility of the government to fund for their re-education. These blue-collar workers usually do not have the resources to pay for a college education. If the government doesn’t help these workers, they simply won’t be able to re-educate themselves for the changing market needs and will fall into poverty. David H. Autor supports this view in his piece for the Journal of Economic Perspectives, “Why Are There Still So Many Jobs? The History and Future of Workplace Automation.” He argues that, due to the rapidly changing dynamics of the AI economy, job displacement will rise significantly if education programs for low-skilled workers does not take place:

…human capital investment must be at the heart of any long-term strategy for producing skills that are complemented by rather than substituted for by technological change.


A Minnesota factory worker with Google Glass 2. Source: Wired Magazine.

Nevertheless, he’s still fairly confident that AI will not completely displace jobs, but rather complement them. Many blue-collar workers such as plumbers, electricians, HVAC installers and others will use AI to become more productive in their jobs, but not necessarily be replaced completely by it. I agree with his view. Microsoft and Google have both released virtual reality goggles that are being tested to aid workers in their day-to-day work lives. The machine tells the maintenance worker where to put the screws on, where he can find the part that’s missing, etc. In fact, Google has already implemented a revamped version of it’s hyped Google Glass product on a factory in Jackson, Minnesota (This is an highly interesting article which I will probably comment on another time. You can find the original article from Wired magazine here). I do not want to dwell on these commendable efforts. Rather, I am much more concerned with the employees of large corporations that perform task-intensive jobs day-in and day-out. Think of the thousands of workers in Foxconn factories building iPhones, or truck drivers delivering merchandise. It is estimate that self-driving trucks, “could threaten or alter 2.2 million to 3.1 million existing U.S. jobs.” What will happen then? A commenter for the previously mentioned MIT article had some truthful insight when he wrote:

The problem is not the technology: it’s the implicit and explicit social and business agreements we have presently in society.

The ultimate problem with job displacement is not so much an issue with unavoidable technological advances that will lead people without jobs. It’s that us, as a society, have failed to properly organize ourselves to fit the needs of the market and put in the required resources into the training and well-being of our workers. Public companies are put under immense pressure to perform, and have put profits over its people (not that it’s a new issue). If we are to avoid a massive displacement of jobs, we need government and businesses to employ appropriate measures to protect its workers by providing them with the necessary education and skills that will enable them to stay competitive in an AI economy It is our duty to use our God-given talents to help others, and therefore the virtue of a good society to provide means for its people to achieve this end.


Corporate Oligopoly

Ah, we enter into a favorite topic of doomsayers and conspiracists. The idea that a few companies will reap most of the profits from a market is far from new: Six movie studios receive almost 87% of American film revenue (boxofficemojo.com), Facebook and Google account for almost 50% of the online ad market and are responsible for 99% of online ad growth, Russia is still controlled by a few oil producers, etc. The list of examples would be endless, and oligopolies aren’t always bad for an economy. They can streamline the production process for goods and services, lower prices for consumers, and provide greater profits to its shareholders.

I strongly believe the AI market will inevitably become an oligopoly (if it isn’t one already), and profits will become even more concentrated in the future. Facebook, Alphabet (parent company of Google), Amazon, Alibaba, Microsoft and Netflix are the leading technology companies in the world. They’re all S&P 500 stocks, have delivered returns much greater than the market, are leading the world in AI implementation and innovating at the fastest rates as well. They also show no sign of slowing down. They have methodically disrupted every industry they have touched -the release of a trademark from Amazon was enough to plunge meal-kit delivery company Blue Apron by more than 30%-, and have digitized many of their processes. They have also concentrated the wealth of these industries among relatively small teams. WhatsApp was bought by Facebook for $50 BILLION and had only 50 employees…50 EMPLOYEES.

Screen Shot 2017-07-27 at 4.11.21 PM.png

Careful there! Each one of these employees is worth $1 billion

Due to a talent shortage in data and AI, these companies compete one another by offering perks and stock options to employees. Startups also frequently do this, as a way to defer salaries to its employees while it starts earning money. Its fine and all, except when these companies grow to enormous valuations and the first few employees hold the majority of the company’s wealth. Amazon still pays its warehouse employees $12 the hour (per Glassdoor), while the company’s valuation is worth $500 billion and its CEO is the richest man on earth (as of July, 2017). A recent article by the Guardian newspaper showed how Nicole, a cafeteria worker for Facebook’s headquarters, still lives in a garage with her family and barely making ends meet. “He doesn’t have to go around the world,” said Nicole. “He should learn what’s happening in this city.” She’s referring to Zuckerberg’s highly publicized world tour that started as his new year’s resolution to “get out and talk to more people.”

“They look at us like we’re lower, like we don’t matter,” said Nicole of the Facebook employees. “We don’t live the dream. The techies are living the dream. It’s for them.” Source: The Guardian

It’s unfortunate cases like Nicole’s that highlight the growing divide between the middle class and the high class being populated by techies. In a new report highlighted by CNBC, a record number of Americans were millionaires in 2016 – there was also a record 50% decline in the people who qualify as middle class, and “One in three say they couldn’t come up with $2,000 if faced with an emergency.” Thus, the corporate oligopoly has concentrated the wealth of the new economy to it’s founders, and the promise that the masses will be liberated to freelance and work on their own thanks to new digital technologies, is shown to be false, except for a fortunate few.

The Rise of Enormous Data

Think Big Data was too big too handle? Enter Enormous Data. Seriously. It’s the new buzzword in the industry.

Stay tuned for updates and I appreciate your comments and suggestions.


Love & Forgiveness

Loving someone does not mean you have to like them. That’s basically the gist I got from CS Lewis’ chapter on Forgiveness in Mere Christianity.

It’s funny how often we think we have to like someone if we are to love them. To think that I need to like the people who cut across me in traffic is tedious at best, unwilling at worst. I can forgive them (at least, in the abstract concept) but I will not feel any sort of sympathy or “warm feelings” toward them..

Good news: That’s ok!

As St. Thomas Aquinas put it, love is not a feeling, it is the active will of desiring the good of the other, for the other. In that sense, we can be assured that forgiveness can be achived without having to actually “like someone.” Still, it would be the Christian thing to do to reach out to the other and desire him/her infinite happiness in heaven, as we all are called.