Is it possible that a robot rebellion occurs or two computers fall in love? Artificial intelligence (AI) raises a series of quirky questions, but in the real world, it involves machine learning as well as deep learning and many other programmable capabilities that we are just beginning to explore. Below we explain what artificial intelligence is, how it works and where it is headed.
What is artificial intelligence?
AI seeks to process and respond to data as any human would. In fact, developers are currently converting a wide range of applications to human-type intelligence.
It is generally classified into three categories, but there is still disagreement about the exact definitions of what artificial intelligence is and, much less, if they are possible.
- Weak: Weak AI (or “narrow AI”) is where the work has been concentrated so far. It is focused on executing a single task, and therefore interactions are limited. Examples include reviewing weather reports, monitoring smart home devices, or giving us answers to general questions that are obtained from a central database (Wikipedia, etc.). Several narrow AIs can be brought together to offer a more complete service: Alexa, Google Assistant, Siri and Cortana are excellent examples, and even the current forms of autonomous cars. Because she cannot think for herself and lacks the ability to understand context, she sometimes provides meaningless answers.
- Strong: Strong AI (or “general AI”) is where we are going. Artificial intelligence understands context and making judgments. Over time, she learns, is able to make decisions, even in uncertain times or without previous information available, use reason and be creative. On paper, they would function as a human brain. So far it has not been achieved, although most believe that this century could be achieved.
- Super: In the distant future, AI can become superior to humans in every respect. They would be able to think for themselves and operate without any human participation. This sounds like a Skynet-like dystopia, with the end of humanity, but it could also be the beginning of an era of innovation.
Artificial intelligence can also be classified based on how it works, which is important when considering how complex an AI system is and its ultimate cost. If a company creates an AI solution, the first question should be: “Will it learn via training or inference?”
- Training: These AIs are designed to learn and improve over time, and to fine-tune your data sets and parts of your processes. Strong and super platforms will be able to do it, but weak AI cannot, since the amount of processing power required is so great that it makes it expensive.
- Inference: Most weak AIs are designed to analyze data and draw conclusions in careful steps, a cheaper and computationally less method. For example, to answer the question “What was yesterday’s game result?” An AI might infer: “I must find data for yesterday’s game results from a list of reliable sports data, comparing them to teams. favorites set in setup and I will report the results by audio. ” While useful, if the answer wasn’t what the user was looking for, the AI has little ability to adapt. A human must be involved to make their responses more relevant.
These definitions are only considered as a general guide to what artificial intelligence is and others may have different descriptions. Anyway, there are examples of current AI that are worth discussing.
Current uses of AI
- Voice assistants: Siri, Cortana, Alex, and other voice assistants are becoming more common, becoming the “face” of modern AI. A growing subset are chatbots, which manage messages on websites and have online conversations.
- Translation: It’s not just about translating the language. It’s also about translating objects, images, and sounds into data, which can then be used in various algorithms.
- Predictive systems: These AIs analyze statistical data and draw valuable conclusions for governments, investors, doctors, meteorologists and, in general, almost all areas where statistics and prediction are important.
- Marketing: These AIs analyze buyers and their behavior, then choose tactics, products, and offers that best suit them. There are currently a number of crossovers between these tools and voice assistants at this time.
- Investigation: Research AIs like Iris search complex documents and studies for specific information, usually at higher speeds than Google’s search engine.
- Awareness: They monitor and report unusual events when humans cannot. One of the most complex examples of this is theft detection, which reports unusual behavior. Along those lines are also autonomous cars, which use artificial intelligence systems to detect dangers and choose the appropriate course of action.
- Editing softwares: These basic AIs look at images or text and locate ways they could be improved.
Where is it going
Neural network expert Charles J. Simon opined on the future of AI, noting that it currently has most of the necessary pieces of strong intelligence, only they still don’t work very well together.
This is a key point in the discussion about what is artificial intelligence and its future. AI is improving, at least that’s the perception, because developers are linking several platforms, although they do not dialogue. For example, Alexa can start your car, but it cannot use the weather conditions to adjust the car’s air conditioning systems. Simon argues that we may have the capacity to do so.
Companies are investing large amounts of money in artificial intelligence, and as long as they’re willing to do it, things will move quickly. But there are still barriers, such as an economic downturn, computational challenges, and even moral and philosophical obstacles, so the road to a Skynet could be a long one.
Is artificial intelligence dangerous?
AIs are long chains of programmed responses and instant data collections, without the ability to make truly independent decisions. That way, evil is definitely off the table for the time being. But that does not mean that human error can do so.
For example, if an AI predicts storms on the East Coast next week, you can send resources and warnings for that area to prepare. But if they appear in the Gulf of Mexico, that inaccurate prediction could have put lives at risk. No one would blame the AI, instead they would look at the data inputs and algorithm settings. Like other types of software, AIs are complex tools for the benefit of people.
At least for now, we know what artificial intelligence is and that it is largely harmless and useful to the world at large. But that could change in the distant future, and at that time we will have to have a serious discussion about how much of our lives we are willing to give to machines.
* Updated by Rodrigo Orellana on July 2, 2020