I’m not talking about SkyNet (Terminator Franchise), or HAL 9000 (2001: A Space Odyssey) but an A.I. (artificial intelligence) that can communicate and interact with us helping us with daily tasks, and even surpass humans in some aspects.
I was recently talking to my dad about this topic, and he said that he thinks we already have AI, and that everyone has some on their smartphones. He was of course referring to electronic assistants, such as Siri, Cortana or OK Google to name a few. This is because back when he was growing up in the ‘70s he remembers a TV show that was telling people about what they can expect technology to produce in the future. In this show it said that in the future, people will be able to talk to their electronic devices, and for it to understand them. This would then trigger a program to open, or for it to reply with something, back in the ‘70s this was advertised as artificial intelligence.
Tomorrow’s World : Yesteryear’s foresight of the future of science and technology.
Now, 40 years into the future, it is fair to say that the goal posts have moved a touch. This is because now we view AI as something that can think for itself and understand what it is doing, rather than just following some code, when it is prompted to. However, some may argue that this is an AI, even if it be a very basic one, as these phones are replying by follow code, or “instructions”, and that this is the same way that humans communicate, as when we talk, our minds are forming words based upon rules that we have, for example grammar, and simple language structures, so if they’re doing the same as us, haven’t we created something that can think like us, just not to the same extent? This threshold is very unclear at the minute.
There is a famous quote from a poem, “to understand the future, you must first know your past” so that’s what I’m going to do.[divider]History Lesson[/divider]
The first man to have ever contemplated the concept of an AI was, you guessed it, English born mathematician Alan Turing (1912-1954), the man who became famous for cracking the German Enigma code, and helping the Allies win World War 2. This was recently demonstrated very well in a film called “The Imitation Game” a film I would highly recommend. He first thought about this when he developed he Enigma cracking machine, and he noticed that his machine could crack 150 million million million possibilities to find the unencrypted version of the message from the Germans, in about 10 minutes, where as if a human sat down and did it, it would take days, if not weeks to crack one message, and even that would only be found by chance.
This got Turing thinking, that if this one machine can do a job so well and so much better than a human, then surely we can develop more machines to do more jobs. On the whole, this theory has been proven to be correct as now we have super-computers, capable of calculating hundreds of Tera-Flops of data each second, much faster than a humans mind.[divider]Progress[/divider]
All aspects of AIs are based on humans, the senses they have, how they calculate, and store information/experiences, their articulation, everything. After all, why wouldn’t you? It’s the best model we have for intelligence, as it the one we understand most. I’m sure that way after we have first come up with AI, we will start to make improvements, but for now, we just want something to show for our efforts, and this will give us the best results.
Something we can use to measure intelligence is the human’s ability to adapt to situations and environments, and also to learn new skills. The reason for how we can do this is currently unknown by scientists, and even though it seems like something that is taken for granted by us. Trying to replicate this on a machine seems to be an almost impossible task so far. If this can be conquered, then I truly believe that AI will be achieved.
Another way for us to measure AI, is the ability to be creative, and to create art and music etc. However we are much closer to completing this over allowing them to learn new skills. There are computers in America developed by a certain scientist that is capable of reading people’s heart rate, and converting this into an “emotion reading”. They then use a specific software to simulate paint brush strokes and creates what they call art. I have to say I have seen some of these drawings and I must say that they look very impressive.
Also, to help create a true AI, we need to replicate our senses for a machine to use, or at least 2 or 3. The main priorities are vision and hearing, and we believe that the other 3 senses aren’t completely necessary, yet. The robot can survive and function without them in short.
We have basically mastered voice, as this is demonstrated by voice recognition software, this is demonstrated well in things like Siri and OK Google, they are both very responsive and accurate, and I think this is a good testimonial to our advances in recreating the human ability to hear.
Sight is proving to be more of a challenge however, as no one piece of software has truly nailed it, or at least not enough for it to be implemented in a stupidly expensive robot. There are lots of basic forms of sight, for example face recognition, or Xbox’s Kinect face log in, however nothing yet has been developed that can detect human emotions etc. a very good example that I can think of, is if two humans are fighting, an AI with the technology today will not be able to tell if the people are play fighting, or genuinely in a fight. This could result in a completely inappropriate reaction from an AI, or no reaction at all, which could be bad especially if the AI is supposed to be babysitting some children for example.[divider]Future[/divider]
AI gas come on leaps and bounds in recent times. One of the greatest demonstrations of this is back in 2011, on a popular American TV show “Jeopardy!” in which a computer, by the name of “Watson” this machine competed against one of Jeopardy’s top competitor, and the machine won, by quite a lot in truth. I think that this shows just how good AI is, and what potential it has, if it can out speed some of the best humans by answering quick fire questions in under 3 seconds. The whole point of machines and computers is to help us with mundane, and boring tasks, like Turing’s Enigma code breaker machine. However, in many aspects these machines have surpassed humans, and will continue to.
I think that he time scale we are looking at before a true AI is born is in decades, not years, and it will require lots of technological break-throughs before any real progress is made. At the moment, we have lots of super computers that are able to be creative in their own rights and to have basic senses. However we have not yet mastered allowing a computer to think for itself, as so far, they are just following instructions, or code, and as a result, when they’re faced with something that they haven’t been programmed to deal with, the struggle, so they do not yet possess that human quality to adapt, and learn as they go. Google specifically have recently had a breakthrough with their new “thought vectors”, which are a new type of way for computers to think, allowing every letter to be code with a corresponding string of numbers. This allows computers to understand what the words you are asking them means, rather than the words just triggering a response. Obviously it is way more complicated than that, however, understandably Google won’t release any father information, to try and prevent too much competition by companies such as IBM and Microsoft. This could offer a huge opportunity for developers to really get advances made in the field of AI.
So no matter how much faster a computer processor may be over a brain, the brain is still superior, at the moment. In the future, who knows? SkyNet is still a thing of science fiction, and hopefully it will stay like that, but we, as humans, can create something extraordinary, even if it is only to help on a domestic scale.