Category Archives: Life

Why NLP is cooler than Rocket Science: Language as a tool for understanding cognition and accelerating the advent of ASI

[<TL;WR>: It mostly says it all in the title.]

In the past two or three weeks, approaching the end of my second semester at the University of Stuttgart, I spent quite a bit of time contemplating what I am learning in a broader context, and trying to understand what it actually is that I want to accomplish.

This challenge was amplified by the fact, that over the past two years, we have seen rise to two companies which I have admired for quite a long time; SpaceX and Tesla. I loved cars and rockets growing up as a kid (and I spent a lot of time drawing my own). Especially the emergence of the F9R (Falcon 9-Reusable) Rocket, which for the first time in the history of mankind tries to commercialize a propulsive landing system, seemed to violently awaken my less abstract engineering spirits, because, lets face it: Propulsive landing is insanely cool.

Falcon 9 CRS-6 Barge Landing Attempt

Now as exciting as that is, it was also a little depressing for me, because upon first glance, my studies of NLP could not be further apart from the kind of hard engineering knowledge that is required in building these incredible machines. I seriously questioned whether my line of studies was really the best way to maximize my personal contribution to the technological progression of humanity.

It took me a while, but I have come to the conclusion that that is not the case. This is not to disregard the enormous importance of “hard engineering knowledge”, e.g. electrical, mechanical, chemical & materials. But as much as those more traditional fields are an important foundation, I think it is fair to say that by themselves they do not generate a lot of progress. It is the contributions from Information Technology that really makes their applications exciting. Take for example the F9R. Yes, fundamentally it is a culmination of exploiting every possible bit of hard engineering knowledge. But there seem to be key contributions from IT without which this would have been impossible (at least given the resources of SpaceX):

  1. Simulation Technology (see this excellent presentation by SpaceX Director of Research Adam Lichtl and SpaceX Lead Software Engineer Stephen Jones on how SpaceX is exploiting sim tech)
  2. Advanced Avionics (to quote the Falcon 9 User Guide: [...] Avionics include rugged flight computers, GPS receivers, inertial measurement units, SpaceX ‐ designed and manufactured controllers for vehicle control (propulsion, valve, pressurization, separation, and payload interfaces) [...])
  3. Advanced Computer Aided Design (Including 3D Printing). (See this video from SpaceX featuring Elon Musk talking about the future of HMI.)

This is the first important observation: What really made the difference was not advances in engineering, but advances in IT. I challenge every reader to argue with that.

My second self-convincing observation (I call it a hope-servation, as in maybe I see what I hope to see) is that we are probably not nearly at the limits of said three areas of contribution from IT to engineering. (Especially processing-power wise, of course, but that is more like an engineering contribution to IT than vice versa.) But this is where NLP comes in. The holy grail of NLP of course is ASI, Artificial Strong Intelligence. Yes, there are more mechanical approaches, that try to neglect the role of cognition in Speech Generation and Processing, but I think there is a trend towards at least incorporating “Statistical Approaches” (Machine Learning), deviating from the purely rule-based approaches proposed by Chomsky.

For a nice introduction into the subject, watch this video by Google:

And this is exactly where I see my contribution to engineering: With the advent of strong NLP, culminating in ASI, there will be an explosion of ever more advanced applications of traditional engineering knowledge. Engineers will be able to literally discuss designs with their computers. Simulators will be able to identify border cases and design very effective test plans on their own.

SpaceX CRS-7 disintegrating after second stage failure.

Events like the Falcon 9 explosion on June 28th 2015 remind us, that human engineering is far from perfect, because the complexity of our machines tends to increasingly exceed our cognitive capabilities. This is why we need artificial intelligence.

 

 

 

Why I will leave Ghana: Telecommuting, the University of Cape Coast, Yandex and Natural Language Processing

As I just published, I am going to leave Ghana in August. I would like to explain the “Why” a little bit more.

Some of you know, that for the largest part of the past year I was very certain that I would stay in Ghana at least until the end of 2014. I went to some length to establish conditions that would allow me to do so: I struck a telecommuting deal with my employer in Germany whilst filling a temporary lecturing vacancy at the University of Cape Coast in Ghana; the combination of which would be reason enough to stay.

Over the past four months though, the bright outlook I saw in staying here transformed into a nightmare of power and internet outages, miscommunication and bureaucracy.

OneUniversity_of_Cape_Coast_(UCC)_crest[1] of my main conclusions is, that I was far too optimistic about promises made to me here. Primarily that goes to the UCC (University of Cape Coast). Prior to initiating the telecommuting agreement with my employer, I made clear to the University that the precondition for my lecturing there would be the availability of a secure work environment for my telecommuting job. It devolved into a bit of a carrot-and-stick situation: I was shown my office in the university, even temporarily holding the keys to it, and continuously promised that I would receive a contract for my work either tomorrow or next week.

Fortunately I learned through a series of happy coincidents that the program coordinator who originally invited me into the university was just a temporary fill-in for the actual guy, and probably didn’t even have the authority to create a contract for me. It is still puzzling me why he didn’t just admit that.

Now I got to know the actual program coordinator. He flat-out admitted that he didn’t have the authority to employ me. That left my dream of staying at the university shattered in pieces, until I had the idea of studying there. They would’ve actually permitted me to do a Master’s in Computer Science, which sounded nice to me. I wanted to stay in Ghana.

That was until I was contactYa logo 250ed by Yandex. Yandex is a Russian Company that operates, among other services, the 4th-largest search engine in the world. They picked up my resume on LinkedIn and invited me to participate in an interview for a C++ Software Engineering position in Berlin. I was really curious, especially because the job would have been about enhancing their maps service. I made it through the challenge tasks and the first interview, but didn’t quite make the second interview. Throughout the interview process though I came to peace with the thought of leaving Ghana for something that may not be as culturally demanding, but far more academically challenging (than studying at the UCC).

After I failed on the second interview with Yandex, I was obviously stuck with a very depressing thought: What if, even after getting a Masters Degree in Computer Science from the UCC, I still wouldn’t have the knowledge to make it through an interview such as the one I just failed?

This is not to discredit the ICT/Computer Science program of the UCC. But indisputably, I do have access to universities in Germany that have far more competence in the field. And also indisputably, Computer Science is one of my core interests.

ims-stuttgart stuttgartSo I started looking for academic “computer sciencey” programs in Germany, and actually found one at the University of Stuttgart that seemed to fulfill my urge for academic depth, especially towards creative algorithms and data structures. The program name is “Maschinelle Sprachverarbeitung”, translated to English  as “Natural Language Processing”.

This is a really important step for me. I had to choose between science and social engagement. After a year of social work in Ghana, I chose science.