Artificial intelligence (AI) is becoming more and more common in our lives. It is no longer limited to certain industries or research institutions; AI is now for everyone.
It’s hard to dodge the deluge of AI content being produced, and even harder to understand the many terms used. But we can’t have conversations about AI without understanding the concepts behind it.
We’ve put together a glossary of terms we think everyone should know if they want to keep up.
An algorithm is a set of instructions given to a computer to solve a problem or perform calculations that convert data into useful information.
The alignment problem refers to the discrepancy between our intended goals for an AI system and the output it produces. A poorly aligned system can perform better, yet behave in a way that is against human values. We have seen an example of that in 2015 when an image recognition algorithm used by Google Photos found photos of black people auto-tagging as “gorillas”.
Artificial General Intelligence (AGI)
Artificial general intelligence refers to a hypothetical point in the future at which AI is expected to match (or surpass) human cognitive abilities. Most AI experts agree that this will happen, but disagree on specific details, such as when it will happen and whether it will result in AI systems that are fully autonomous.
Read more: Will AI ever reach human-level intelligence? We asked five experts
Artificial Neural Network (ANN)
Artificial Neural Networks are computer algorithms used within a branch of AI called deep learning. They are made up of layers of interconnected nodes in a way that the neural circuits of the human brain.
Big data refers to data sets that are much larger and more complex than traditional data. These datasets, which far exceed the storage capacity of household computers, have helped current AI models perform with a high degree of accuracy.
Big data can be characterized by four V’s: “volume” refers to the total amount of data, “velocity” refers to how fast the data grows, “truth” refers to how complex the data is, and “variety” refers to the different formats in which entering the data.
The chinese room thought experiment was first proposed by the American philosopher John Searle in 1980. It posits that a computer program, no matter how apparently intelligently designed it will never be conscious and will not be able to actually to understand his behavior as a human does.
This concept often comes up in conversations about AI tools like ChatGPT, which appear to exhibit the characteristics of a self-aware entity, but really just present output based on predictions from the underlying model.
Deep learning is a category within the machine learning branch of AI. Deep learning systems use advanced neural networks and can process large amounts of complex data to achieve higher accuracy.
These systems perform well on relatively complex tasks and can even exhibit human-like intelligent behavior.
A diffusion model is an AI model that learns by adding random “noise” to a set of training data before removing it, then assessing the differences. The goal is to learn about the underlying patterns or relationships in data that are not immediately obvious.
These models are designed to self-correct when they encounter new data and are therefore particularly useful in situations where there is uncertainty or when the problem is very complex.
Explainable AI is an emerging, interdisciplinary field concerned with creating methods that increase users’ confidence in the processes of AI systems.
Due to the inherent complexity of certain AI models, their inner workings are often opaque and we cannot say for sure why they produce the outputs they produce. Explainable AI aims to make these “black box” systems more transparent.
These are AI systems that generate new content – including text, images, audio and video content – in response to prompts. Popular examples are ChatGPT, DALL-E 2 and Midjourney.
Data labeling is the process of categorizing data points to help an AI model make sense of the data. This includes identifying data structures (such as image, text, audio, or video) and adding labels (such as tags and classes) to the data.
Humans do the labeling before machine learning kicks in. The labeled data is split into different datasets for training, validation, and testing.
The training set is fed to the system for learning. The validation set is used to verify that the model is performing as expected and when parameter tuning and training can stop. The test set is used to evaluate the performance of the completed model.
Large Language Model (LLM)
Large Language Models (LLM) are trained on massive amounts of unlabeled text. They analyze data, learn the patterns between words and can produce human-like responses. Some examples of AI systems using large language models are OpenAI’s GPT series and Google’s BERT and LaMDA series.
Machine learning is a branch of AI that trains AI systems to analyze data, learn patterns, and make predictions without specific human instructions.
Natural Language Processing (NLP)
While large language models are a specific type of AI model used for language-related tasks, natural language processing is the broader AI field that focuses on the ability of machines to learn, understand, and produce human language.
Parameters are the settings used to tune machine learning models. You can think of them as the programmed weights and biases that a model uses when making a prediction or performing a task.
Because parameters determine how the model will process and analyze data, they also determine how it will perform. An example of a parameter is the number of neurons in a certain layer of the neural network. Increasing the number of neurons allows the neural network to handle more complex tasks, but the trade-off is increased computing time and cost.
The Responsible AI movement advocates developing and deploying AI systems in a human-centered manner.
One aspect of this is to enshrine AI systems with rules that ensure they adhere to ethical principles. This would (ideally) prevent them from producing output that is biased or discriminatory or could otherwise lead to harmful results.
Sentiment analysis is a technique in natural language processing that is used to emotions behind a text. It captures implicit information, such as the tone of the author and the degree of positive or negative expression.
Supervised learning is a machine learning approach that uses labeled data to train an algorithm to make predictions. The algorithm learns to match the labeled input data with the correct output. Having learned from a large number of examples, it can continue to make predictions as new data is presented.
Training data is the (usually labeled) data used to teach AI systems how to make predictions. The accuracy and representativeness of training data have a major impact on the effectiveness of a model.
A transformer is a type of deep learning model mainly used in natural language processing tasks.
The transformer is designed to process sequential data, such as natural language text, and figure out how the different parts relate to each other. This can be compared to how a person reading a sentence pays attention to the order of the words in order to understand the meaning of the sentence as a whole.
An example is the generative pre-trained transformer (GPT), on which the ChatGPT chatbot runs. The GPT model uses a transformer to learn from a large corpus of unlabeled text.
The Turing test is a machine intelligence concept first introduced by computer scientist Alan Turing in 1950.
It was designed as a way to determine if a computer can exhibit human intelligence. In the test, computer and human output are compared by a human judge. If the outputs are indistinguishable, the computer has passed the test.
from Google LaMDA and OpenAIs ChatGPT reportedly passed the Turing test – though say critics the results reveal the limitations of using the test to compare computer and human intelligence.
Unsupervised learning is a machine learning approach where algorithms are trained on unlabeled data. Without human intervention, the system examines patterns in the data, aiming to discover unidentified patterns that can be used for further analysis.