My favorite game (and an upcoming AI game I’m thrilled for!)

As of now, my favorite game is GTA 5 (Grand Theft Auto V).

The premise of the game and its franchise is about playing criminal characters in an open world storyline. It is notorious for being one of the most popular, revolutionary, and innovative games on the market, whilst containing the most violent, sexual, and drug-ridden characters/plots ever!

With that being said, the reason that it’s been regarded as so innovative is due to its extensive lead in the following areas of gaming:

• World size
  • The world is huge. To walk across it from one end to the other, it’ll take around 2 hours. I’ve played the game for a few months now and there are most certainly locations I haven’t been to. It takes around 10 minutes to drive across in a fast car. The terrain varies from city, to mountains, to ocean, to lake, to small towns, to military bases, etc. It is huge.
• World density & interactivity
  • There is so much you can do aside from the side missions and storyline. Exploring is a huge thing. Many Youtubers make entire videos on just driving around hidden GTA 5 locations, finding cool stuff. But there are a plethora of activities. From skydiving, to flying jets, to driving submarines, playing chess, to doing triathlons, golf, tennis, invading military bases, discovering items, solving clues engraved into terrain, discovering easter eggs eg. UFOs, etc. etc. And the world is so dense. It’s filled with content and amazingly intricate. The ocean is even filled with marine life and coral. Everything is thought out. There are even other little things that will blow your mind: http://www.ign.com/articles/2013/09/30/100-little-things-in-gta-5-that-will-blow-your-mind. It is a living, breathing world.
• Graphics
  • The new GTA games are always leaders in the visual design and graphics. GTA 5 is no exception. The graphics on the PS4 and Xbox One are beautiful.
• Characters/story
  • The characters and storyline/plot in GTA games are always addictive, well casted, and super fun. GTA 5, again, is no exception.

GTA 5 trailer: https://www.youtube.com/watch?v=mzHTUpe_b0Q

It’s not just GTA 5, it’s all the previous GTAs too. They have been as successful and revolutionary, and are landmarks in gaming. They have pioneered the open world game market.

GTA 5 was given 11/10 by many major tech and game reviewers. Seriously.

I think this might change, soon! A small indie game dev company recently announced a new game called No Man’s Sky. The hype surrounding this game has been immense. The game’s premise is all about exploring. The team has Math, CS, Economics, and Biology & Physics PhDs on their team working on algorithms for procedural content generation. The No Man’s Sky universe is multiplayer, and it’s huge. It contains 18 quintillion planets that have all been generated by their patented and state of the art algorithm. Everything down to the terrain, animals, alien monsters, features, etc. are procedurally generated. When the game launches, 99.9999% of these planets will have never been explored. You can explore them. There are some missions and storylines in the game but they are weak – the main premise of the game is exploration with others in the multiplayer universe. There is one main objective: to get to the center of the universe, where supposedly something insane happens. Again, though, 18 quintilion planets is insane. Just think about it. If the game sells 50 million copies (which is around what GTA 5, the most popular game, sold), the average player would have to explore twenty billion planets for the whole universe to be fully explored.

No Man’s Sky: https://www.youtube.com/watch?v=nLtmEjqzg7M

Minorities in Silicon Valley

This blog will be short as I am working on my final project.

As someone who is very much in the Silicon Valley community – the industry’s biggest problem is undoubtedly the (lack of) opportunities of minorities and females. Facebook only hired single digit black people last year.

There have been numerous cases that deserve attention, most notably Ellen Pao and her lawsuit with Kleiner Perkins, but also within other companies like Github and Google.

The industry is aware and retroactively working on making itself more inclusive to minorities and females in tech, celebrating gay pride and discussing these issues loud and clear. Sam Altman, president of Y Combinator, himself has discussed how woman tech co-founders have a harder time raising funds.

Question is: why aren’t there many woman in tech? Is it due to stereotypes (that coding is more of a man’s job) or lack of inclusion. I think it’s definitely a hybrid. The former is being solved with many organizations celebrating, funding, supporting, and raising awareness for woman who code. The latter seems to be some sort of hidden evil within the industry. Hopefully, however, as time goes by, the efforts made by the community will continue propagate forward and ignorant views against minorities/females in tech will be pushed out. This article is one example.

Data policy discussion

This is a topic I find very interesting. Let me address some of these issues people have with Facebook, then make a more general comment.

1. Keeping Your Data Forever, Even If You Leave

This is fine to me – I actually feel uncomfortable with the fact that my information will literally be completely lost if I left Facebook. If I’ve made a rash decision and find that 30 years later I’d like to obtain some of the information (for nostalgia’s sake or whatever), it would be sad if the data was literally gone from the face of the Earth.

2. Telling Your Friends What You Buy Online

This one I have a problem with, but Facebook fixed it. I should be able to control what my friends see about me on an extremely granular level. I can understand if the data was posted to Facebook itself, but any transaction made on another site deserves my control over. They fixed it with Facebook Connect, and are now actually putting more emphasis on privacy. These days, privacy sells.

3. Tracking Your Movements Across the Web

I like this, and explain it under the “Why I’m fine with my data being minded” section.

4. Using Your ‘Likes’ In Ads

I don’t necessarily like ads, but I appreciate them (because without them Facebook wouldn’t make money and I wouldn’t be able to use it). I discuss more about this in the following sections.

5. Forcing You to Make Your Data Searchable

Also fine with this. You’re putting your information on a public site, what do you expect?. It involves your name and profile picture and other super basic info that doesn’t expose anything really important or potentially harmful about you. (Unless you are a criminal, of course).

6. Using Facial Recognition Software to Spot You in Photos

This is fine to me and allows me to easily access photos of me that I didn’t take (but participated in) and/or forgot to upload. It allows me to go back on more fun photos. Only problem is: it automatically tags you and inserts it into your timeline. This annoys me, because it’s external information that now my friends see without my permission. I can go into my timeline to hide it, but that’s more effort on my part. There might be a setting to change this that I am not aware of.

7. Giving Your Data to the Government

I’m not a criminal, I don’t care.

Underlying theme:

As soon as I make bad decisions in life, I’ll face the consequences. If I undergo any bad criminal activity, I deserve to have my private data looked at. At the end of the day, security (people’s lives) > privacy – it’s a reason why I support the NSA. The government nor Facebook cares about any swear words or other stuff via personal chat – they wouldn’t even have time for it.

Why I’m fine with my data being mined:

Security > context > privacy. Security is the most important: people’s lives are more important than people’s privacy. Then: context. I would prefer to have a personalized experience with content that I find relevant, interesting, or helpful (example: Facebook Feed and Google Now Predictions). I would much rather know if my flight is delayed in the easiest and fastest way possible than be worried of Google mining my emails (which again; are just to improve my experience). This is a case where I disagree with Apple/Tim Cook and find Apple’s products/services lacking.

Contingencies:

I am in control of who sees what. I don’t care if it stays on the system or if I am shown ads – these are just computed recommendations, not human curated. At the end of the day, I’m just one piece of Mathematical data in a sea of data. But the site needs to be super secure, because as soon as someone looks at my data with the goal of knowing about me and breaching my life somehow vs. making money it poses a problem. So I only share information with sites that are secure and constantly improving security.

Implicit bias (and what it entails in the context of AI)

I took an online test that told me if I had an implicit bias between African Americans and European Americans. After about 10 minutes of answering speed questions – it informed me that I have an implicit bias towards African Americans.

But what does this mean in the context of AI?

I’d like to relate it back to the article by Turing we read last week. One main and underlying argument was that machines fail to consciously think independently and synthesize truly new information – they have no bias, and rather carry on the bias of the person designing or writing it. They are not in control of their minds.

Perhaps this statement is true, but even then, is inherently a false argument for the debate at point. Because, if implicit bias is something that exists, maybe we are just the same? Maybe our creators implemented bias into our systems, and we aren’t fully in control of our thoughts. Just some speculation.

xkcd comic thoughts

This comic portrays a man who has fallen for an online partner, but suspects her of actually being a chatbot. He then confirms this suspicion after she fails to solve a Captcha in an online couples testing tool/software.

It is obviously unfortunate that the man had to endure months of real emotions only to discover the person on the other end was/is not real (and hence no opportunity for an actual relationship), but there is a bigger picture here: trust. If chatbots could pass the Turing test just like this one, it entails a lot of danger. Malicious people could setup bots to fool people into thinking they are not who they actually are, and hence these bots would be a tool for stealing sensitive information (and things like that). They pose a real danger.

Day 2 programming excercises

Meme: 

Random Circles: 

Seurat: 

OpenProccessing code

Link: http://www.openprocessing.org/sketch/206301

void setup() {  //setup function called initially, only once
  size(250, 250);
  rectMode(CENTER);
  background(0);  //set background to black
  colorMode(HSB);   //set colors to Hue, Saturation, Brightness mode
}
 
void draw() {  //draw function loops
  background(0);
  if(mousePressed == true) { //add some interaction
    rect(mouseX, mouseY, 50, 50);
  }
  else {
     ellipse(mouseX, mouseY, 50, 50);
  }
}

With this code, I draw a 50×50 white circle (ellipse()) on the screen in the mouse’s position when the mouse is not clicked, and a square (rect()) when it is clicked.

Response to Turing article (on consciousness)

I definitely agree with most of Turing’s responses towards arguments against human-like AI. The theological objection is one that we can certainly disregard due to a lack of proof and general cloud of contradicting beliefs amongst the believers of different religions – the lack of clarity and “derived”/descending views indicates that this conjecture will most likely stay false. I would like to further expand on this topic below.

A popular and underlying theme that present in this argument against Turing and others (and one that I feel Turing failed to comment on) is that humans seem to have some form of continuous “mind” and/or “soul” – abstracted away from what is in reality the brain. As humans, it is rather easy to fall into the sand-boxed mindset that our mental attributes (personality, emotions, memories, thoughts, etc.) aka our “definition” is actually not physical and is some separate entity inhabiting our body. Even if people consciously refute this, they most likely subconsciously fall back into this state (mostly due to the distractions that life gives us). When you think about it, and even though I’m less familiar with this subset of biology, our personality, emotions, fluid thoughts, etc. are in reality just chemical reactions, electrical impulses, etc. – all powered by or located in the brain. It’s just very sophisticated and developed. Just like code and Math is most likely the medium for AI in our world, chemicals or electrical impulses may be the same tools that our creators used to create us, while they are in fact built from something even more sophisticated (at the time). It’s hard to comment on exactly what that could be – in the same way that imagining living in a four dimensional world poses.

A good way to escape this sandbox is to ask: what if I was dead? What would it feel like? It would feel like nothing. Your brain isn’t working, and **you are your brain**. If we look at AI – how could we know what it is doing (whatever it may be – learning by adjusting weight vectors based on gradient descent) isn’t a form of thinking itself? Our creators could have potentially once had (or have) the same mindset as us. To them, our thinking could be simply artificial.

Here’s an even better way to think about it: let’s say we are in a simulation, and our superiors decide to kill someone. First: how do you feel about killing a piece of Math/code in a simulation? Nothing… So why would it be different for our creators? And why would it be different for our creations? How do you know Math/code isn’t consciousness?

It’s hard to elaborate what I mean, but as you keep reading, you may start to realize that our thoughts, emotions, memories, personality, etc. are nothing special but an output of some function or reaction. They make us function somehow. It’s more of a gradual realization than something you can tap into immediately. Our coded creations could be living in the same sense that we are, and have that same sort of “soul” or “mind”. I think our definitions of “thinking”, “consciousness”, “living”, etc. are what potentially traps us into this mindset.

Basically, we could just be code. Our feelings, thoughts, etc. could just be code. They are real, sure, but not “real” in the same light we hold them. They are potentially not “realer” than our Math/code. We think we are meaningful but we may be meaningless. And our thoughts, emotions, etc. can prove this otherwise – our brains may just be good tech. You could almost consider the brain a separate entity on its own. The brain thinks for itself and about itself and feels different things in different states eg. depression. Think about it for a second. Imagine removing the brain and connecting it to a lever which has different options eg. excitement, depression, anger, etc. It’s just a **state that is what it is**, yet it entails so much more for us. Does this maybe make my argument slightly clearer?

In a sense, you can zoom out, and view yourself as just a thing that produces stuff like emotions and inhabits those produced emotions.

**So now**: couldn’t the machines fall into the exact same mindset that we do? That’s where my argument concludes.

Hard to explain my thoughts, but this is why I do not believe in the consciousness and theological arguments.

AI stuff

3 that I found interesting:

b & c. As for driving in California or Cairo, it’s definitely possible and has been done by Google (Google self driving cars). I’ve read up on this before, and the logic works the following way: the computer needs to be trained first. So it has to have some sort of camera attached to the top so it can see the road ahead. Then, it needs to track the driver’s wheel motion and movements. Basically it needs to “observe” the driver. Pairing the wheel motion/movement with the processed/quantified image of the road will serve as labelled, training data for supervised learning. After a while (and a lot of observing and learning), the car should have been trained well enough so that it finds a correlation between the wheel motion/movement and road direction/path so that it makes few errors on the training data itself, and it can now drive by itself. It’s probably less accurate in Cairo because the maps aren’t as detailed than in California.

h. I find this one interesting. I think that it’s possible with a combination of Machine Learning (specifically Neural Networks) and NLP algorithms, but not yet. You essentially give the system a bunch of funny stories and unfunny stories, labelled as funny and unfunny. This will serve as training data for the neural network. The actual input data will not be the story, however. Sentiments paired with concepts and entities extracted from the story with NLP algorithms should be fed forward through the neural network as input, and it can be trained with backprop to optimize on training data. I think, over time, it’ll learn what kind of words, pair of words, etc. and sentiments make a story funny. The problem is piecing together an actual story – what story would it write about? This is probably the biggest challenge. It may know whether something is funny or not (and we can teach it that), but generating a new story would be incredibly difficult. On top of that, some things are simply much too complex eg sarcasm may be difficult for a program to learn, and may cause conflict in training data. With the right techniques though, it could be possible in the future.

About me (First Post)

Hey, I’m Rohan (http://rohankapur.com). I’m a 16 year old student iOS, Backend and Web developer living in Singapore. I was born in Australia, and moved to Singapore when I was 7 – my parents are from India.

I’m a rising junior at the international school I attend in Singapore (United World College of South East Asia). I started programming in Objective-C when I was 11, and have made several apps. In 2013, I won a student scholarship to attend Apple’s WWDC. With my most recent app and startup, Contra (getcontra.com), a platform for millennials to share their opinions on different social issues, I expanded into backend development (specifically with Go, Python, and Node.js) and web development. Contra has been featured and promoted by Apple in the App Store, and we’ve managed to amass an active user base who use the app everyday.

I’ve been interested in Machine Learning and AI for a while now. I’ve completed and passed Andrew Ng’s Machine Learning course on Coursera and am now pursuing Geoffrey Hinton’s (the guy who invented backpropagation) Neural Networks course. I am interested in Philosophy and its link with CS and specifically Artificial Intelligence. Now that I have a good deal of users, content and events/data, I am working on NLP and data mining algorithms to provide the user tailored recommendations and also to understand and quantify the user’s opinions and perspectives on different entities, concepts, and topics which can be used for market research. This will most likely be my final project for this course.

Outside of CS, I am interested in entrepreneurship, especially since I am treating Contra as a startup and have raised angel funding. I also enjoy playing the guitar and piano. I hope to attend Stanford as an undergrad and work in Silicon Valley.