Artificial Intelligence involves different fields. We have Machine Learning, where the systems are able to "learn" by themselves. Probabilistic Methods, used in cases of incomplete or uncertain information. Search and Optimization Methods, where we do smart searching among many possible solutions. And we also have the Statistical Methods, using, for instance, logistic regression to build a classification system.
 |
| AI-based system dashboard - Photo by Stephen Dawson on Unsplash |
In the last ten years, lots of AI algorithms were applied in countless business from different areas of expertise. Why AI is just now a hype sort of thing and not before? Did you know that the AI algorithms started to be developed in the fifties? Why just now they are being applied? Why they took so long to make use of AI? Well, there is a simple answer to that. There were not enough data to feed the AI algorithms, especially the machine learning ones. At that time, the amounts of data produced by people and computers were very small. But with the growth of technology and the exponential use of the internet, we are now capable of producing millions and millions of records.
Ok, but why we need so many data? To properly train a neural network, being it a convolutional network or not, we need lots of data to achieve a better fit of our model to produce good results. With few data to train with, our model will overfit to the training data and will not be plenty generic (will not generalize) as we wish. In the next posts of this blog, I will describe in detail these problems as well as teach how to build neural networks.
We are constantly facing a big shift in our daily lives thanks to continuous technological evolution. Nowadays, we see systems capable of predicting our needs even if we do not tell them what we want. These systems are able to collect and analyse data about us and find patterns leading to user demands. Off course, we have some ethic, security, and other issues about how we are being, in some way, spied to give enough data to the called smart systems used in the industry. You can say “but I do not share anything I do not want to”, however, I can tell you, certainly you are sharing lots and lots of data that is being collected every second and analysed by AI algorithms to give you smart predictions and suggestions adjusted to your behavior. By simply using any google tool, like the google chrome, everything you type in and click is being recorded to give you ads tailored to your needs. That is the most simple example I can give you regarding how we have many ethic and security issues with AI. Most of the people send constantly data to these AI algorithms without even knowing they are doing it.
Is it illegal? I am not sure because I am not a lawyer, but I can bet it is not illegal because they tell us everything in the section of terms and conditions that we always read, oh wait, we do not. We always accept and do what we need to do because life is too short to read this kind of stuff. So, we can not claim they are spying on us when we are the ones allowing them to do it.
On the other hand, AI has a huge power to bring growth in several areas we, humans, still need to better understand. Artificial Intelligence can be seen as a technology producing significant, and positive impacts in our everyday lives. AI normally relates to an artificial creation of human-like intelligence. Thus, we expect the technology being able to learn, plan, perceive, reason, or process natural language and talk to us. These features allow the AI technology to be a huge socioeconomic opportunity.
We need to think and find the best algorithm to use for each use case. Thus, we must know all the available possibilities in the AI to build better and more adequate AI systems.
Comments
Post a Comment