Artificial Intelligence But Not Social – Nation World News

Turing Prize winner (considered the Nobel Prize in computing) Geoffrey Hinton said a few months ago that artificial intelligence is going to solve everything. But is it so? Is Artificial Intelligence Really Smart?

When we talk about Artificial Intelligence (AI from now on), we should not forget that we are talking about computer program (code) which is designed, developed and maintained by people so that it Let the program influence and influence the people. AI is therefore created by people. It is true that AI allows you to do new things; We relate differently, we communicate differently, etc. but on the bottom The concern is another: that one group of humans has advantage and control over another group of humans., Machines have no emotion, feeling, desire, purpose, or autonomy. However, humans do. And this is where public power must rise to control what is done and what is not.

The areas in which society can obtain intelligence are very broad. They can be described from the point of view of civil service, increased job opportunities, better security, access to public information and transparency etc. And, of course, from a more business point of view. And it is that many sectors of economic activity see opportunity in this data economy. And thats where Hintons phrase we quoted at the beginning makes sense. Is everything solvable? Whats more, do we want it to be able to solve everything?

Till date, Machines have not naturally demonstrated their ability to understand and abstract, It is in fact the basis of human intelligence. This is what sets us apart from other species. So we are currently talking more about extended intelligence than artificial intelligence. In addition, and more importantly, we have problems with interpretability and transparency. When we use neural networks, for example, which are autonomic (and quite opaque in terms of how and what they learn), it is difficult to understand how much weight is given to each variable to predict something. Do we rely on this ambiguity to predict the possible development of diseases? In other words, even if we dont know why an algorithm has learned to prioritize some variables over others, do we rely on it to make decisions?

Perhaps we should start softening these absolutist discourses, and bet on simpler models, which tend to be less successful, but at least we know what and how they predict. This area of AI is called eXplainable AI (XAI). It suggests that if we want an algorithmic society, we need to know which variables are relevant, report confidence intervals, and describe what it has learned. That is, actually understanding the magic that happens is based on looking for patterns within the algorithm.

Otherwise, we would create dark boxes in what Frank Pascal calls the Black Box Society. The book The Black Box Society: The Secret Algorithms that Control Money and Information, by the above author, introduces the ethics of technology from three perspectives that should be considered when we build algorithms: (1) that The resulting result must satisfy a series of rules, policies, principles, etc.; (2) evaluate the results obtained by these algorithms; (3) Include clearly and implicitly the shared values in the society in which they will be introduced.

as you can see, This is not an exercise to regulate technologies. But what can we humans do with it?, There is still no artificial intelligence law in any country. Yes, many agendas and plans. The power of data is clear. Not so much the ethics and the social and inclusive perspective. Whether the power brought by these figures is in our hands or not. And the thing is, AI is more human than artificial. And it is about expanding human capabilities rather than artificially altering them. He

Read more:

Artificial Intelligence But Not Social - Nation World News