Artificial intelligence: AI Terms Simply Explained

Artificial Intelligence

John McCarthy, one of the founding figures of AI, described it as “the science and technology of creating intelligent machines,” serving as a domain of exploration for researchers and engineers. Today, AI primarily refers to computer systems exhibiting intelligence.

Weak/Strong AI

 AI differs from human capabilities by excelling predominantly in a single task, characterizing it as weak or narrow AI. In its specialized domain, current AI systems often outperform humans.

General/Sturdy AI

The quest for an AI possessing human-like intelligence, capable of diverse tasks, remains unfulfilled, termed General AI. It is sometimes referred to as strong or true AI, though the conceptual boundaries can be hazy.

Super-AI

Should General AI surpass human capabilities to an extent that it exceeds conventional understanding, it becomes artificial superintelligence. While some foresee it as a potential threat, others hold hope that super AI could resolve profound global challenges.

Manufacturing AI

AI can be developed through two distinct approaches:

  1. Good, Old-fashioned AI (GOFAI): This approach, prevalent until the late 1980s, aimed at achieving strong AI by logically combining individual terms representing our knowledge about the world.
  2. Machine Learning (ML): Presently favored in AI research, particularly deep learning, ML creates computer systems that learn from data, continuously optimizing their performance, rather than relying on explicit, rule-based programming.

(Artificial) Neural Networks

Inspired by a basic model of the human brain, artificial neural networks consist of interconnected layers of nodes exchanging information. This typically includes an input layer, hidden layers, and an output layer.

Deep Learning (DL)

Deep learning involves neural networks with multiple hidden layers, and it gained prominence around 2012 when such networks won the ImageNet competition for image analysis. This technology has fueled recent advances in AI, particularly in image recognition, autonomous driving, and deep fakes.

The rise of deep learning is facilitated by faster processors, dedicated AI chips, such as Google’s TPU, and extensive data for training.

Supervised Learning

In supervised learning, AI is trained using labeled data, meaning it is provided with pre-annotated training examples to learn from, like recognizing objects in images.

Unsupervised Learning

Unsupervised learning is a promising area of AI research where AI is given large datasets without explicit labels, and it autonomously discovers patterns within the data, often uncovering hidden correlations.

Conclusion :

In the words of AI luminary Yann LeCun, AI resembles a cake: most of it represents unsupervised learning, supervised learning is the icing on top, and reinforcement learning serves as the cherry.

If you like it, then keep reading the top university in Delhi, NCR viz. Geeta University’s blogs!!!