Before talking about artificial intelligence, I think it is important to try to define what we understand by intelligence, as we did in class, because "if we cannot define it, how do we think to create it?" Intelligence can be perceived by some as the act to recognize our reflection in a mirror, or the act of communicating, understanding or being able to perform complex operations.
"...the capacity for logic, understanding, self-awareness, learning, emotional knowledge, reasoning, planning, creativity, critical thinking, and problem solving. More generally, it can be described as the ability to perceive or infer information, and to retain it as knowledge to be applied towards adaptive behaviors within an environment or context. Intelligence is most often studied in humans but has also been observed in both non-human animals and in plants. Intelligence in machines is called artificial intelligence, which is commonly implemented in computer systems using programs and, sometimes, specialized hardware."
Plato used to say that babies know what a horse is before knowing it, and then he gives it a name, while Aristotle thinks that you only know the world through experience. That´s what they call the Top-Down & Bottom-Up approaches. (Symbols vs Senses). Two different approaches to the learning process: trough experiencing it or learning something without experiencing it.
We talked a lot about different models of intelligence like the "Brute Forcing" which means learning something by trial and error until you find the way to solve it; or "Neural Networks" which can derive conclusions from a complex and seemingly unrelated set of information to generate a concept or hypothesis by trying to find characteristics between them; or the "Compulsive Neural Networks" which are mostly used for image analysis; or "The Expert Systems", which can emulate the decision making process of a human being to solve a problem or make a decision; or the "Case-Based Knowledge" which helps us to solve a problem through the knowledge given by previous problems solved in the same way.
Most of these AI procedures must be fed in a certain way by previous information through a machine learning process. A process in which we teach the machine with a lot of different possibilities so that it can understand the subject we are talking about. For example, if we want it to differentiate between a dog and a cat we will need to feed it with a lot of pictures of the two animals, defining which is which so in the future it can do it manually. These sets of information are called Datasets, and we learned how to search them in different popular websites, and also how to start analyzing them through BigDataML.
Whit my group we tried to learn about how to teach a machine to detect emotions in language, but deeper than that, we wanted to teach it to detect sarcasm in conversations, something that requires a deep comprehension of language and context to define.
We found a dataset of the most used words in the English language, and what someone did to discover emotions trough this dataset was to give different values to each word, related to 7 basic emotions like anger, happiness, hate, fear, and others. So in that way, the machine could detect which word was more related to which feeling, and by that, give a certain score to a sentence including those words. We didn´t get to the point of discovering how to discover sarcasm in a sentence, but we found some investigations done around that, that used the same scoring system of the other dataset to try to find incongruence between words in the same phrase, so in the same sentence someone uses a positive quote, related to a negative one, there could be a possibility of sarcasm in language.
It was interesting also to debate about the ethics of AI since it is something in what we do not always think of. We think of AI as something we just create and leave there to work by itself, but it is something that has a lot of responsibility since everything that the machine does depends on the instructions, protocols or limits you have added to the code. If the machine kills someone is a decision taken before by someone else, it´s not a decision taken without premeditation. And it happens the same with the rest of what we design because we might have created it for something "Good", but it might be used by someone in a different way, like a Knife, which doesn´t kill by itself.
The week vas good, but the learning had to be very superficial because of the lack of time.
This site was created with Mobirise