The ABCs of AI

Industries from healthcare to weather forecasting and supply chains are embracing the efficiency and automation provided by AI solutions. As AI continues to penetrate the market, there are multiple terms that can help your brand navigate and better understand artificial intelligence, machine learning systems, and training methods.

Artificial intelligence is a branch of computer science broadly focused on creating artificially intelligent systems and machines that are capable of mimicking human intelligence and the human mind. Artificially intelligent systems usually ‘learn’ how to do so by processing huge data sets, identifying patterns that they can model decision-making on in real-time.

Also known as Strong AI, this term describes a machine that can truly think like a human being, essentially replicating human intelligence by solving problems that it has not been specifically trained to work on via its cumulative experience.

This is the science fiction version of AI often seen in movies and on television; a kind of artificial intelligence that could, in theory, pass the Turing Test (named after Alan Turing, who called his version of the concept “the imitation game” in 1950). AGI does not yet exist, though Google DeepMind’s AlphaGo made headlines as a possible breakthrough in this area in 2016.

This type of AI could have enormous problem-solving potential and far-ranging use cases, but it also comes with a corresponding difficulty level and large ethical concerns about the creation of such a system have been raised during AI research. The next level of advancement would be artificial superintelligence (ASI), a system that is superior to humans.

A subset of artificial intelligence where AI assists the way humans work instead of replacing them entirely. AU is focused on helping humans make better decisions with tools designed to enhance their cognitive abilities and assist in removing bias and human error.

A type of AI solution that can simulate conversation, such as a chatbot or virtual assistant like Apple’s Siri, Amazon’s Alexa, or the Google Assistant technology. It is powered by natural language processing (NLP), which focuses on enabling machines to not only process human language but understand the nuances it contains.

A training method in AI systems that teaches computers to process data in a way that is inspired by the human brain. Due to this design, deep learning models can recognize complex patterns in the data they are trained on and use those patterns to produce accurate results.

A philosophy that prioritizes human understanding of the impact, accuracy, outcomes, and biases of AI models. XAI is an approach that helps users have confidence in the results of machine learning algorithms and is especially important when it comes to building trust and confidence among those using AI models. XAI can also help brands craft a responsible approach to the development and integration of AI across their organization.

An umbrella term that broadly refers to any AI that generates something ‘on its own’ including images, text, or video. ChatGPT and its image-generating equivalent DALL-E fall into this category.

A class of machine learning and a common framework in generative AI. GANs consist of two neural networks: a generator and a discriminator. The neural networks work together during the training process and both become more skilled, as a result. GANs are a popular approach in generative AI frameworks because these two neural networks are especially useful in the generation of multimedia assets.

A specific type of large language model (LLM), created by OpenAI and popularized in ChatGPT-3 and ChatGPT-4. GPT is trained on large text-based data sets. These models produce responses to inputs that are not only rational but textually relevant.

A large language model is also part of the generative AI subset. LLMs are a type of algorithm that uses deep learning and enormous data sets to generate human-like text by studying the text that has been provided and predicting the next word or symbol. LLMs are the technology behind most chatbots and voice assistants.

Also known as weak AI or specialized AI, this is the term for an AI system designed for one specific task, such as predictive analytics or speech recognition. Email spam filters and even autonomous vehicles like self-driving cars fall into this category.

A subfield that bridges linguistics, computer science, and artificial intelligence. NLP is focused on interactions between computers and human language. It looks for specific ways to program computers to analyze large amounts of natural language data, so it’s possible for a computer to organize documents and also process and extract information based on the nuances of language inside the documents.

NLP is key to many AI technologies that assist marketing teams because of its utility in analyzing social media posts, customer reviews, and support tickets.

A machine learning model based on the neural structure of the human brain. It organizes interconnected nodes into layers — just as the human brain does with neurons — and adjusts the weight of the connections as it processes data.

A machine learning model that can most easily be described as 'learn by doing’ where the system is rewarded for making the correct choice until it learns to make the correct choice on its own.

Robotic process automation (RPA) is software technology where users create software robots, or “bots”, that can learn, mimic, and then execute rules-based business processes.

A subset of both machine learning and artificial intelligence, supervised machine learning is uniquely defined by its use of multiple pairs of input objects and desired output values to train an algorithm.

Used in machine learning, transfer learning is the application of a previously trained model on a new problem. In transfer learning, a machine uses the knowledge gained from a previous task to improve generalization about another.

An approach in machine learning where machines look for similarities in unlabeled training data and learn from the data itself rather than being instructed by humans.

To learn more about AI and machine learning, visit the Sitecore Knowledge Center.