Its a long blog but do read it till the last to know the complete scenario.
After the invention of the wheel, the internet was probably the best invention of mankind in recent times, but suddenly it is changing in unfathomable ways by the implementation of Artificial Intelligence all over the internet and its implementation in robotics. This blog discusses the potential future of modern AI and how it will change our lives in the coming future.
Ever heard of the world's most complex board game called Go, well, it was invented 2500 years ago in China. There is no brute force method to win this game because it has 10 to the power of 170 possible combinations; as an interesting fact, no atoms in the entire observable universe is 10 to the power of 80. Recently Google's deepmind alpha Go beat the 18 times world champion in one to one match. Strategies developed over the period of 2500 years were learned by this AI in a matter of weeks. This AI was trained on millions of game sample from real humans and developed its own strategies to win the game. Right now alpha Go's is the best Go player in the world but technically it's not a human. This event shows that there is more and more non-human knowledge about everything. The way AI is developing I'm pretty sure that it will learn all the rudimentary things we humans do. But wait, isn't it great to do all the rudimentary task, this seems fine but when you look closely that if AI continues to develop in a similar manner consequence can be dire. There is a term called 'Singularity' when we'll reach singularity humans will become just as important as chimps. We'll discuss this in later sections.
So, what are the current trends and capabilities of today's AI? AI is divided into two parts narrow AI and general AI. Narrow AI is just good at one thing like Alpha Go can only play Go, but what if it is given with some other much simpler game, will it perform the same or the better? As we know that it has learned a much complex game so it will win the simpler game but things don't work that way. Alpha Go AI will definitely lose the game because it doesn't even understand what game it is playing if it sees a similar pattern it will respond accordingly. Here comes the general AI, it is not trained to a specific thing rather it is trained to understand everything. Building general AI is hard as of now there is no general AI which can understand everything, but it doesn't mean that we can't reach that point. Given enough time AI will be able to understand, right now to learn anything AI uses a large amount of data so to get a dataset which contains the knowledge of entire mankind is tough. But what if AI learns to create programs which can create data and then learn from it. The way AI is evolving this rapidly is because electronic chips are much faster as compared to brain neurons and it also has higher bandwidth as compared to the brain.
So back to the Singularity, it is the point when an AI becomes smarter than the entire mankind. It'll be a dangerous thing because once that happens you can't control it. On the other hand, it'll provide solutions for seemingly unsolvable problems. Will this be the end of every problem we have, No, I don't think so, a time when AI will understand everything will come very soon but it won't be able to solve all our problems. Let's say an AI is tasked to solve the world hunger and it proposes a solution in which everyone dies, it will not hesitate to do so because this may be the only way to solve it. So here comes the thing which makes us better than an AI, it is called wisdom. Intelligence takes is bein the previous data, produces an output, make some mistake, learns from it and it keeps going on whereas wisdom is applying the correct knowledge in the most appropriate way. Wisdom is being able to see beyond the information and apply knowledge to different things in a productive way.
AI has brought up the biggest and most pressing question will it be pleasant or hateful could it even be either of those things in the first place and more importantly is that even necessary. Sure, a super-intelligent AI would have all the information in the world but then again so do you and I so do all the people in the world with very radical views who's to say that a super-intelligent AI wouldn't adopt these views as a stone and then decide to execute on those instead of what we had planned. Sure it's possible that a super-intelligent AI could improve upon itself using the knowledge gained but then again it isn't pulling random improvements out of thin air it's not just going to unite quantum mechanics and general relativity with the raw mathematics. What if some of its upgrades needed to experiments to be run what if it needs more information? Is it just going to make it up this leads into us into what if learned to lie and then decides to lie to us about its accomplishments for its own selfish reasons it's also possible that the progress of computing will slow down just like it already is in case of airplanes, in past 100 years planes design barely changed, like many things there's a threshold that is hard to get past but if we do get past this barrier the upsides could be tremendous it's almost obvious that superintelligent AI has the capability of making humanity billions of times better than it is today but on the flip side, it could also be used as a weapon just think like I said before if this thing is running for a week and has access to all of the world's information it could makeover 20,000 years of technological progress you let this thing run for six months and you could potentially have 500,000 years or even more of technological progress imagine if this got into the wrong hands and imagine the repercussions it would have when you google something not only are you giving out data and telling a potential artificial intelligence what you're thinking you're also telling it how you're thinking. AI is potentially the last invention of humanity this developments that could possibly turn the human species into immortal godlike figures is coincidentally the same technology that could also cause the downfall of humanity many people are resistant to the AI because they fear it may have some of the negative traits that we humans do it may be greedy it may lie it may be selfish it may be rebellious it may be short-tempered the power of superintelligent AI is there patiently waiting to be found, but the question is do we actually want to find it?
Comments
Post a Comment