What is simple and weak fundamental

Strong AI, weak AI - what can artificial intelligence do?

Artificial intelligence is basically divided into strong and weak AI. While weak artificial intelligence already reaches or surpasses human intelligence in defined sub-areas, strong artificial intelligence would be on at least the same level as the human brain in all areas.

Weak artificial intelligence

As weak artificial intelligence weak AIornarrow AI) are systems that focus on solving specific application problems. The problem solving takes place on the basis of methods of mathematics and computer science, which are specially developed and optimized for the respective requirement. The resulting system is able to optimize itself.

Weak AI systems work reactively on a superficial intelligence level and do not gain a deeper understanding of problem solving. Weak artificial intelligence is primarily geared towards fulfilling clearly defined tasks and does not vary the approach to problems. Instead, the weak artificial intelligence falls back on methods that are made available to it for problem solving.

All systems in existence today fall under the category of weak AI.

In the scientific and philosophical discussion, systems are sometimes referred to as weak artificial intelligence, which only seem intelligent, but behind which there is no artificial intelligence in the narrower sense. In the majority of cases, these are purely rule-based systems.

Examples of weak AI

Weak artificial intelligence is already used in many areas of everyday life. Some examples are ...

  • Character and text recognition
  • Image recognition
  • voice recognition
  • Individual control of advertising
  • Automated translation
  • Expert systems (e.g. deriving recommendations for action based on a knowledge database)
  • Navigation systems
  • Auto-completion and suggestions for corrections in searches

The majority of these systems come from large software manufacturers (Google, IBM, Microsoft, Apple, etc.). However, more and more smaller providers (startups) are emerging that develop weak intelligences for solving certain problems or tasks.

There is currently particularly strong growth in, for example, intelligent chatbots that take over customer support. In the B2B area, accounting is automated and optimized by weak AI, as is the inbox and the back office.

Strong artificial intelligence

The goal of a strong artificial intelligence (also Super intelligence, or English: strong AIorgeneral AI) is to achieve or exceed people's same intellectual skills.

A strong artificial intelligence no longer acts only reactively, but also of its own accord, intelligently and flexibly.

As of today, it has not yet been possible to develop a strong artificial intelligence. And the discussion as to whether the development of such intelligence is even possible continues. The majority of researchers now agree that strong AI will be developed, but they disagree on when this will happen.

However, a period of 20 to 40 years is considered realistic.

Properties of a strong AI

Despite all the speculation and uncertainty, there is already a basic consensus among most researchers and scientists that a strong intelligence must have the following properties in order to be considered such:

  1. Logical thinking skills
  2. Ability to make decisions even with uncertainty
  3. Ability to plan and learn
  4. Ability to communicate in natural language
  5. Combine all skills to achieve an overarching goal

On the other hand, the question of whether strong AI can achieve its own consciousness and what role empathy, self-knowledge, memory and wisdom play in connection with the aforementioned properties remains open. It is speculated, for example, that a superintelligence will not have real empathy, but could only simulate it to achieve faster or better goals.

If you want to delve deeper into the topic of super intelligence or strong AI, you should get the book "Superintelligence: Scenarios of a Coming Revolution" by Nick Bostrom.

/ 17 Comments / by Julian Moeser