AI Terms
It is difficult to keep up with the growth and capabilities of AI, and the extent to which it has been developed is likely the tip of the iceberg.
Understanding some AI terms can give us a clearer picture of what exists now, and what could be coming in the future.
Artificial Intelligence: a term coined in 1955 by Stanford Professor John McCarthy as "the science and engineering of making intelligent machines." To go a step further, AI is the theory and development of machines that can perform tasks that usually require human intelligance.
Types of Artificial Intelligence
AI attempts to emulate human functioning, so one criterion for determining its type is its capability to do so (Capability-based). AI can also be sorted for functionality, which shows how it applies learning to interact with its environment and process data (Functionality-based).
Capability-based AI
Artificial Narrow Intelligence (ANI)
AI tools designed to autonomously complete specific commands or actions. These tools do what they are told to do, and have a limited (or narrow) range of competency and cannot learn independently. ANI sometimes uses neural network algorithms and machine learning. All current AI falls under this category.
Artificial General Intelligence (AGI)
AI that has the ability to learn, think, and perform at a level similar to humans. Multifunctional. This AI is still in progress, but could be built from existing technologies and AI models like ChatGPT.
Artificial Super Intelligence (ASI)
In addition to the multi-function of AGI, ASI will have the memory, processing speed, and capability for analysis that could help it perform at a higher level than humans.
Functionality-based AI
Reactive Machines
- AI that reacts to requests and stimuli.
- Task-specific
- Has no memory; cannot learn
- Examples: Email spam filter, IBM's Deep Blue, programs that give suggestions based on consumers previous purchases or search history
Limited Memory
- AI that stores past data and makes predictions using it
- Learns from historical data to make decisions.
- Examples: Almost all current AI tools, including chatbots, self-driving cars, image recognition software, virtual assistants like Siri and Alexa
Theory of Mind
- Conceptual AI that perceives and picks up on emotion, and predicts future actions based upon it.
- Far off; difficult to develop because of the nuance required to read emotional cues.
Self-Aware
- Hypothetical AI that is so evolved that it is self-aware
- Understands and evokes emotions and has emotions, beliefs, and desires of its own.
- Researchers simultaneously strive to create this AI and fear the consequences of its creation.
A finite procedure or set of steps followed to sove a problem or complete a task.
Everyday examples include following a recipe, solving a long division problem, tying shoes, finding a book at the library
In AI, algorithms can allow machines to learn, analyze data, and make decisions based on what they have learned.
Extremely large data sets that can be computationally analyzed (mined) to reveal patterns.
The process of sorting through large data sets to discover patterns that can be used to solve problems or improve processes.
A subset of machine learning that uses the human brain as inspiration for teaching computers to process data. Instead of using an algorithm that completes a specific task, it can learn from unstructured data.
Examples: virtual assistants, facial recognition, self-driving cars, social media algorithms, wearable sensors
A form of artificial intelligence that generates new content from scratch including text, code, audio, images, and video. This is possible because its neural networks have been trained on large amounts of data.
Possible outputs: essays, blog posts, computer code, ad copy, press releases, digital art
Examples: Chat GPT, Bing Image Create, DALL-E, Synthesia, OpenAI Codex, Boomy AI
A form of text-based generative Al (e.g. ChatGPT) that is trained on an enormous amount of text so that it can predict and create a given sequence of words. This capability allows the model to "understand" inquiries and replicate human language in a largely coherent (if not always accurate) way.
The development of computer systems that can learn and adapt without following explicit instructions. To do this, they use algorithms and statistical models to analyze and draw inferences from data patterns.
A type of machine learning process using computer networks made of layers of interconnected nodes. Neural Networks mimic the way neurons in our brains interract.
CompTIA AI Advisory Council Glossary
A University Leader's Glossary for AI and Machine Learning (Inside Higher Ed)
Machine Learning and Higher Education (Educause Review)
A Generative AI Primer (Jisc National Centre for AI)
Artificial Intelligence and teh Future of Teaching and Learning (US Office of Educational Technology)