From Weak to Strong Artificial Intelligence
Artificial General Intelligence (AGI) is the next Artificial Intelligence (AI) phase coming soon after exiting the weak phase. No doubt, we are still living in the weak AI era: Artificial Narrow Intelligence (ANI). ANI presents every mobile app, face recognition, search engines, and speech recognition tools. However powerful we feel about our smartphones, we are still smarter.
Every ANI tool limit is the task it performs. Google Search, for example, performs complex computations based on highly sophisticated algorithms that even genius minds can’t bear. But it’s boundary is the single task of “searching.” Even Google Translate is an ANI example.
However, researchers are making their way into another form of AI, the strong AI. Represented simply by AGI, there are enormous obstacles that scientists must overcome to reach their goal. AGI symbolizes human-like machines that can understand and reason their surroundings.
The more researchers dive into the AGI, the more they get lost. It’s difficult to build a machine that feels its environment, fears risky situation, and makes decisions based on its perception. Therefore, AGI remains ambiguous so far.
Challenges Facing Artificial General Intelligence
According to Gary Marcus the Professor of Psychology at New York University, CEO and Founder of Geometric Intelligence, and author of multiple popular science books, deep learning isn’t sufficient. Marcus insists that various subjects such as “cognitive sciences”, “developmental psychology”, and “developmental cognitive neuroscience” must be researched and linked to AI industry to achieve the next phase. Moreover, Marcus expresses the significance of perception, common sense, reasoning, and planning to achieve AGI.
Nevertheless, challenges facing AGI so far are:
- Granting machines “common sense” and “creativity”.
- Computers performing instructions without a notion of “will” to do something.
- The challenge to reach “multitasking” AI systems.
- Supporting machines with “adaptability” to work in new situations.
- Applying to other tasks what a machine learns to perform for one task.
- Anticipating actions same as a human mind.
- Figuring out how people reason and create consciousness.
- Machines must understand “abstract” notion to resemble human minds.
“Starting Soon” – Subscribe to our blog to receive updates directly to your inbox.