Artificial intelligence (AI) is the science and engineering of making intelligent machines, particularly computer programs. It is related to the task of using computers to understand human intelligence, but AI does not have to confine itself to biologically observable methods. Alan Turing's work, "Computing Machinery and Intelligence," in 1950, questioned the possibility of machines thinking. Turing's "Turing Test" test, published in 1950, remains an important part of AI history and philosophy. Stuart Russell and Peter Norvig published Artificial Intelligence: A Modern Approach, which delved into four potential goals or definitions of AI, dividing computer systems based on rationality and thinking vs. acting. AI is a field that combines computer science and robust datasets for problem-solving, including sub-fields of machine learning and deep learning. OpenAI's ChatGPT has marked a turning point in AI, with generative AI now focusing on natural language processing and learning the grammar of software code, molecules, natural images, and other data types. As AI's use in business continues to grow, discussions about ethics become critically important. IBM stands within the conversation around AI ethics, and IBM Watson Orchestrate and IBM Watson Assistant are two AI platforms that can help enterprises harness the power of AI.
Weak AI, also known as Narrow AI, is focused on specific tasks and is responsible for most of the AI we use today. It is capable of robust applications like Siri, Alexa, IBM Watson, and autonomous vehicles. Strong AI, consisting of AGI and ANSI, is theoretical, with a machine having an intelligence equal to humans and self-aware consciousness. While strong AI has no practical examples, researchers are exploring its development, with examples from science fiction like HAL in 2001: A Space Odyssey.
Deep learning and machine learning are sub-fields of artificial intelligence, with deep learning being a sub-field of machine learning. Deep learning is a neural network with more than three layers, which automates feature extraction and enables larger data sets. It is considered "scalable machine learning" and can leverage labeled datasets, supervised learning, or unstructured data in its raw form. Machine learning, on the other hand, relies on human intervention to learn and determine the hierarchy of features, requiring more structured data. Deep machine learning can ingest unstructured data in its raw form and automatically determine the hierarchy of features, allowing for more interesting scaling options.
Generative AI is a deep-learning model that can generate statistically probable outputs from raw data, such as Wikipedia or Rembrandt's works. It has been used in statistics for analyzing numerical data, but the rise of deep learning allowed it to extend to images, speech, and other complex data types. Variational autoencoders Vases were the first deep-learning models to achieve this cross-over feat. VAEs opened the floodgates to deep generative modeling by making models easier to scale. The future of generative AI is models trained on a broad set of unlabeled data that can be used for different tasks with minimal fine-tuning. Foundation models, trained on large, unlabeled datasets and fine-tuned for various applications, are driving this shift. Foundation models are predicted to accelerate AI adoption in the enterprise by reducing labeling requirements and enabling highly accurate, efficient AI-driven automation. IBM hopes that foundation models can eventually be brought to every enterprise in a frictionless hybrid-cloud environment.
Artificial intelligence (AI) systems have numerous real-world applications, including speech recognition, customer service, computer vision, recommendation engines, and automated stock trading. Speech recognition uses natural language processing to process human speech into written format, making it accessible on mobile devices like Siri. Customer service involves online virtual agents answering frequently asked questions, providing personalized advice, and cross-selling products. Computer vision, powered by convolutional neural networks, enables computers to derive meaningful information from digital images and videos, providing recommendations. Recommendation engines use past consumption behavior data to develop effective cross-selling strategies, making relevant add-on recommendations during checkout processes. Automated stock trading optimizes stock portfolios without human intervention, making thousands or millions of trades per day.
You must be logged in to post a comment.