Days 12, 13 – My vision on AI

AI (Artificial intelligence) is a dynamic field of study, dealing with problems which, if tackled the right way, can change the way we see computers right now.  I consider AI such an amazing subject because it evolves on multiple fronts – algorithms, data science, games etc.

I am personally an algorithms enthusiast – I like designing time efficient algorithms for the hardest P problems and versatile heuristics for NP problems (I usually go for techniques like dynamic programming, greedy algorithms and Divide et Impera + some advanced data structures – segment trees, randomized cartesian trees etc.). To test my skills I usually compete in algorithms contests (which are usually very fast paced contests) – my Codeforces Profile. For me it’s the empathy of getting an “Accepted” verdict that keeps me running sometimes, other times is just for fun’s sake + this helps me train for on-site CS olympiads. I also love discovering and understanding complex insights when dealing with those problems (they are some kind of brain teasers, they sharpen your mind).

Enough about that, I think that after the next major breakthrough in AI (maybe a new optimization process mimicking something obvious which nature already does by itself, other that Neural Network, or maybe theoretical CS will make a breakthrough first, resulting in a few new techniques, ready to be applied in AI on real data) the way people precept the relation human-computer will totally change, I don’t have any bet on when that’s going to happen, I’m otherwise sure it’s going to.

If you were to ask me, AI should for the moment rely on observations from nature to design new amazing techniques:

• Understand the human brain
• Fully understand evolution
• Use what they’ve learned

It may sound like it’s a biologist’s job to do those. NO! They should cooperate. with people who work in AI to plan and prioritize what to do next.

Day 11 – Technology versus Nature

Technology versus Nature

I think that humans can balance nature and tech, but the general problem with is the consensus that “I’m one man/woman, there is 7 billion more, do you think I am going to change anything if I change?”. In part I agree with that way to ration, the problem arises when we bring this in the context of mass psychology. It’s all about the Bandwagon effect – whose immediate consequence is that nobody is going to change before a vast number of other people change (which is a paradoxical situation). Also speaking about willpower, people tend to be lazy when it comes to important changes (this is pure human nature). We appear to be condemned continue destroying our planet while being conscious about doing it. The only reliable way to change something is basically using force (which is not good either) – like strong laws (which unfortunately also means restrictive laws) and mankind to enforce those laws. Introducing a new law meant to stop destroying nature (like trade it for technology) can’t be done in one night, it requires years to get to it’s final form without any protests from the citizens. I agree with what the article says, we are caging ourselves and there is not much to do about it (I’m not pessimistic, but I’m still waiting for a brilliant idea that is going to make a change – what about paying people, in one form or another, to save our planet?)

Day 10 – Game Review

Every now and then I use to play computer games. A few of my favorite titles are Crysis (and it’s sequels, from the shooter genre), Assassin’s Creed and Mirror’s Edge (from the action-adventure genre) and, after all, I’m a huge fan of real-time strategy games, one of the best ones being Company of Heroes (COH) in my opinion.

Company of Heroes is a 2006 real-time strategy video game developed by Relic Entertainment.  A standalone expansion, Opposing Fronts, was released on September 25, 2007. A second standalone expansion, Tales of Valor was released in April 2009.

(Wikipedia)

My arguments in its favor are conceptually very simple:

• Vehicles and units don’t take damage based so much on the health counters (the situations on which that happens are rare and usually get boring – like waiting for one minute for two 5 member squads to fight each other at a range of 10 feet), but rather based on critical hits. Basically what that means is that skilled players can use attrition warfare combined with hit-and-run attacks to their advantage. Basically the combat mechanics are interesting (although unfair sometimes, speaking about the other player getting lucky).
• Unlike in other strategy games like Red Alert 2 it’s more about the quality than the quantity (there exists a population cap controlling how many units can you have, regardless the amount of resources you have). Having a large amount of units can be subject to artillery fire.
• Micromanagement of tanks and light vehicles is extremely important (a squad of riflemen can easily put down alone an armored recon vehicle if the vehicle’s owner doesn’t constantly move it to avoid grenades), unlike in let’s say Panzers.
• The models are extremely realistic and campaign missions historic – The single-player campaign puts the player in some of the major American operations during the Battle of Normandy.
• The demands for producing units is carefully chosen to balance out factions (although some factions are harder and more intensive to play right).
• You basically need to use strategies, not just tactics in order to win.
• Maps are not 2D, fortifying a hill can lead to dominance.
• The large community of players online and its expansions COH: Opposing Fronts and COH: Tales of Valor, each adding new factions and units (and brand-new campaigns).
• The learning curve is also a part of the game – understanding the way more units can be used and what situations they are appropriate for is just part of it, understanding the sets of abilities or playing in tandem in a 2v2 match are just as hard and deep.
• Predicting what your enemy is up to is a definite game changer.

A few downs would be that it requires a high-end computer to play with satisfactory graphics and the length of a typical online match being of ~40 minutes.

Day 10 – LaTeX induction proofs

Day 9 – Data Science for social good

I have found quite a few such projects (and summer programs related to it) – link1, link2. Let’s take as our example “Identifying Fraud & Collusion in International Development Projects“, ran by The World Bank Group. Quoting from their full description description:

Every year, the World Bank Group lends over $30 billion for aid and development to more than 140 countries. However, some analysts estimate that developing countries risk losing billions of dollars each year as a result of fraud and corruption. Misappropriation of funds is most likely to occur at the contract bidding and awarding stage, when bribes, collusion, and other practices can divert money away from projects.

Basically it’s a program about preventing bank fraud and corruption at very high level (though their concern are the small money “leaks”).

The problem with corruption is that its almost invisible for the moment it happens, small amounts of money being completely irrelevant from one account’s perspective, though that happening repeatedly means extremely high loses. A very complex system of detection needs to be used to detect that or more elaborate actions (like using multiple accounts in order for the authorities to lose track of the money being involved). Supervised AI, trained properly with appropriate examples under fitting infrastructure for such detection would be an interesting idea to implement and could revolutionize world economy as we now know it.

Day 8 – Fairness

Unfortunately, capitalism means some jobs getting more attention than others. A mathematician works just as much to get his major as somebody doing CS. The industry wants people who can get code working (even though it might not be that much of a breakthrough for mankind in any sense). Fairness is not always possible in those conditions – it’s about the commercial side of things – good coders are hard to find and are vital for the company’s income, talented mathematicians being even harder to find, but in some employers’ views unnecessary. Programming is not taught in school because at first it may sound not general enough for people who want to pursue philosophy for example.

Contrary to what can be done about gender-bias, there is almost nothing to do, because every private company governs itself and they know exactly what they want. Forcing some laws in this direction would probably not please syndicate members, so that is not an option either.

Day 7 – Computer Science History Museum

Today we visited the Computer Science History Museum. I liked it, though it was the densest museum I ever got to see (and one of the bests at the same time). Everything was so well documented, that I literally spent most of my time reading exhibit names and explanations and trying to make chronological, factual and competitiveness relations between different computers or computer companies. I liked how they had movies about important figures in computing and interactive stuff (like logical gates or the flip-flop circuit in the digital computing section). I was most impressed by Seymour Cray, who revolutionized supercomputing, his designs being 100 times faster than those of the competitors, by using brilliant boolean logic insights. They also had an Enigma machine (which I always wanted to see in person) which reminded me of the humongous effort the Bletchley Park team lead by Alan Turing had to do in order to crack it (also one of my favorite movies – The Imitation Game).

Day 6

Tech shift: More women in computer science classes

Our brain is such an intricate deterministic (but quantum speaking stochastic) system. There are still thousands of processes we don’t fully understand (even though we already scratched the surface in that direction) – gut feelings, eye accessing cues or simply “feeling like doing something”. In a similar manner a man’s and a women’s minds are pretty different at subconscious level. Along some lines somebody may evolutionary argue that the activities a man and a women are somehow “made for” are predetermined. I’m going to counter that by the fact that evolution means crossover and mutation, means inheriting what the species has best to offer. Humans are social (and that’s why women associate CS with “boredom” and “typing” rather than “problem-solving”), rational creatures who can very easily adapt. By the means of that I argue that a man can be inclined to a “girly” activity or vice versa.

In order for the industry to prosper you are going to need representatives of both genders, so that the group can get the best from all sides.

Encouraging women to apply for jobs, to attend lectures or to go for majors in CS is in my opinion a very good thing until the point at which you get to disadvantage men from getting their dream come true, so the “gender-neutral” approach is going to best fit my vision (the 2012 University of Michigan study has confirmed that going the other way around is not the right way to go).

Day 5 – Monday

7 Controversial Ways Facebook Has Used Your Data – Myself I am not that worried about any of those, but I don’t find “secretly” adding very forcing stuff to the Facebook terms and conditions the right thing to do for two reasons:

1) Almost nobody reads those.

2) Even for those who do, the way they present things is often obscure and people are probably going to skip over the important rights Facebook reserves.

3) The remaining 0.5% of people who read the terms thoughtfully probably will make a compromise in order to just use Facebook even if some of their inner rules are violated.

Also “changing the rules of the game while playing” (what Facebook usually does) is not fair play. I find the way Facebook plays is violent, also just half telling the truth about their intentions for some of the terms (like in the Help Center section).

Day 4 – IAT

I took the Age IAT. Even before finishing the test, by the way they were settling the tasks I understood that they want to see which association fits you better Old = Bad and Young = Good or the other way around, by seeing to which version you respond faster and more accurately, this way ranking  your beliefs. In my opinion, the existence of a version which feels more natural to you shouldn’t necessarily mean you are socially or ethically biased. This kinds of tests may tell us how our subconscious works at a high level.