General (Strong) AI (AGI)
- [Picking Wildflowers - Leopold Franz Kowalski]
- Overview
Strong AI, also known as Artificial General Intelligence (AGI), is a theoretical type of AI that aims to create machines with human-like cognitive abilities. The AGI hypothesis is the philosophical position that a computer program that causes a machine to behave like a human would also give the machine a mind.
AGI machines would be able to learn, reason, solve problems, plan for the future, and adapt to new environments. AGI could potentially perform tasks at a human level, and could be applied in many fields, including healthcare and education.
Narrow AI (ANI), is limited to specific tasks and computing specifications. For example, an algorithm that can identify images of cats and dogs is a ANI system. AGI refers to machines possessing generalized intelligence and capabilities on par with human cognition. Unlike ANI systems, AGI would have open-ended abilities to learn, reason, and adapt to unfamiliar environments.
Some say that AGI could be possible within the next five to ten years. Others say that it could be both imminent and distant, and that we'll likely need to solve more specific problems first.
Please refer to the following for more information:
- Wikipedia: Artificial General Intelligence
- The Concept and Characteristics of AGI
Artificial General Intelligence (AGI), also commonly referred to as strong AI or deep AI, is basically the hypothetical intelligence of machines. It is the concept of machines imitating or imitating human thinking or human behavior with the ability to learn and apply this method/intelligence to solve any type of various problems.
Technically, the term AI refers to a model created to solve a specific problem or provide a particular service. What is powering things like ChatGPT is AI. It is learning how to do chat better but can’t learn other tasks.
By contrast, the term AGI refers to software that’s capable of learning any task or subject. AGI doesn’t exist yet - there is a robust debate going on in the computing industry about how to create it, and whether it can even be created at all.
AGI can mimic human reasoning and intelligence to understand a problem and its context and solve it. It has a greater and broader ability to mimic human understanding and intelligence, and remains similar to human intelligence in specific situations.
As of now, researchers are just trying to make deep AI possible. The possibilities and implications of this more powerful and capable AI will be enormous. Thanks to this, AI machines will have greater autonomy and be able to perform multiple tasks with precision, in addition to greater decision-making power through experimental knowledge in various fields.
With comprehensive knowledge and cognitive computing capabilities, AGI can be considered to understand, interpret, and act in a way that is indistinguishable from humans in terms of performance. Researchers have not yet been able to achieve robust AI, but estimates suggest we will be able to achieve it by 2040. Many people wonder if AGI is possible.
- AGI and The Human Brain
Strong AI, also known as artificial general intelligence (AGI), refers to machines with general intelligence and capabilities comparable to human cognition.
Unlike narrower weak AI (ANI) systems, strong AI (AGI) will have an open-ended ability to learn, reason, and adapt to unfamiliar environments. It implies human mastery of fluent skills (such as language, logic, planning, and creativity) that transcend any one domain.
The human brain can do many amazing things, and it may be the most complex and capable creation we've ever seen. But it's also full of flaws. It processes information very slowly, and it learns very slowly compared to computers. Instead of repeating what we already know how to do, we should focus on the tools that can help us. We want humans and machines to cooperate and do things they can't do on their own, and that's what weak AI (ANI) is for.
In theory, then, anything humans can do, AGI can do. We don't have AGI in the world yet. First, the Moravec paradox makes us strive to replicate basic human functions such as vision or movement. (Although image and facial recognition AIs are starting to learn to "see" and classify.) Also, at the moment, AI can only do the few things we've written, and it's clear that AGI is still far away. It is believed that to achieve truly powerful AI, we need to make our machines conscious.
AGI is basically developed in accordance with the AI framework theory of mind, which corresponds to learning human perceptions, needs, emotions, beliefs and different thought processes in different contexts.
Since the human brain is the basic model for building this AI technology, researchers and scientists working on this AI project face too many challenges. A lack of understanding of every aspect of human brain function makes it more difficult.
- Beverly Hills, California - Civil Engineering Discoveries]
- AGI Examples
In short, any task that a human can do can be done by AGI. Technically, it has all the potential of the human brain. It can solve any problem or task in any field, be it music creation or logistics - all potential actions that a human can perform.
This includes the following intellectual tasks:
- Generalize knowledge and apply it to different situations: Humans learn from experience. They take lessons from various experiences and apply them to other situations they encounter. This would be an example of strong artificial intelligence.
- Use the knowledge and experience gained to plan for the future: Another ability of humans is to use their life experiences to plan for the future. As they encounter more experiences, they can use them to make plans and drive the future. In narrow AI, machines must rely on humans to program actions. These machines cannot plan for the future.
- Change and adapt to the environment as things change: General AI machines will be able to adjust as they encounter situations. AI in the narrow sense can only respond to variables programmed into the algorithm. General AI can make decisions on the fly.
- Reasoning ability: Unlike narrow AI, general AI will be able to reason. A general AI machine will be able to examine a situation and determine a course of action, even if it is beyond what humans have taught it.
- Solve a puzzle: Of course, AI algorithms have competed and won video games and chess competitions. These successes are examples of AI following patterns and procedures. There are some confusing issues where success is not currently defined. When machines can solve puzzles of this nature, general artificial intelligence will be realized.
- Show common sense: Another very human attribute is common sense. Common sense is sometimes necessary when machines cannot rely on programming to get answers. Weak AI has no common sense. To be on par with human cognitive abilities, machines must exhibit common sense.
- Consciousness: Machines need to be conscious and self-aware to achieve AGI.
- Beyond math equations: Narrow AI proves in many ways that many of the problems we solve as humans are just mathematical equations. Machines will have human intelligence when they can go beyond mathematical equations to solve general problems.
- Identify needs and emotions: AGI is also able to read the needs, emotions, thought processes and beliefs of other intelligent entities. This is called the theory of mind-level artificial intelligence. This type of artificial intelligence has nothing to simulate or replicate, but machines that truly understand humans.
- AGI vs Generative AI (GenAI)
Generative AI (GenAI) is a type of AI that can generate new content, such as text, images, music, or other data, resembling human-created content. It often uses models like Generative Adversarial Networks (GANs) and Transformer models. GenAI is transforming industries today with its specialized capabilities.
GenAI is a currently implemented subset of AI that focuses on creating new content based on learned patterns from specific datasets. It can be called narrow AI (ANI) because it can only imitate human capabilities and perform simple, specific tasks. However, by simple, we mean more straightforward than the ones performed by AGI, which basically has to be a machine with human cognition.
When GenAI generates new data, it doesn’t realize what it’s actually doing because it lacks proper understanding or reasoning abilities and generates outputs based on statistical patterns from training data. In contrast, AGI systems would possess genuine understanding and reasoning abilities, allowing them to realize their actions.
- AGI: Stepping Stone to Artificial Super Intelligence (ASI)
In order to make ASI a reality, key technologies that need to be further developed include: advanced machine learning algorithms, especially deep learning, capable of complex reasoning and generalization, significantly improved computing power, powerful natural language processing (NLP) to understand and respond to nuanced language, improve the ability to learn and adapt on the fly, and develop AGI as a stepping stone to surpass human intelligence in different fields. In essence, AI systems can not only perform specific tasks and can think and reason like humans in a variety of situations.
