Understanding Terminology

A special neural network model that can convert one type of input into another type of output. Google first used the term when it developed a more efficient neural network for translating English into French in 2017. Transformers are a key component of natural language processing applications such as LLMs and DALL-E.

This is the information used to train machine learning algorithms. It is able to analyse large amounts of data (text, figures, images or sound) and use them to complete tasks. ChatGPT, for example, uses text data that is available on the internet.

In the context of AI, a prompt is an instruction given to a chatbot or other model to obtain a specific reaction or response. Prompt engineering is the technique of using precise and detailed instructions to obtain a highly differentiated response.

A key element in a machine learning algorithm. Put simply, parameters are a set of functions that the AI model uses to fine-tune its ability to achieve the desired result. In general: the greater the number of parameters, the more powerful the model.

This is a research and development laboratory for artificial intelligence that has set itself the task of developing AI for the benefit of humanity. It is the creator of the GPT and DALL-E models.

Ensures that the website content for each region and language is fully customised to the area in question, improving search results by region and/or language. In this case, localisation is essential.

These consist of HTML elements that are invisible to the user.However, this metadata is seen and used by search engines to assess the relevance of websites for the requested search.

This is the modern approach to machine translation used today. Based on artificial neural networks, NMT can predict the likelihood of a sequence of words, modeling entire sentences in translation. NMT can produce human-like translation quality for high-resource languages.

An area of artificial intelligence in which algorithms use data to search for and process linguistic connections and sequences between the words, sentences and paragraphs in both written and spoken human speech.

An area of artificial intelligence in which a computer algorithm searches for patterns in the provided data and then uses new data to complete a requested task. The program “learns” from the information it receives without the need for further instructions.

Machines that can replace functions that normally require human (or other biological) brain power.

Software that has the same capabilities as the human brain in a wide range of functions. This program does not exist yet but it is often seen as the ultimate goal of AI research.

This is a process in which human expertise is combined with machine skills to achieve high-quality results. In the language industry, for example, human specialists edit machine translations to ensure that they do not contain any errors or inaccuracies. Humans can also intervene to evaluate AI models and provide feedback for training.

A type of artificial intelligence that is trained on huge amounts of text data and is therefore able to recognise the patterns and connections between words and sentences. It uses this knowledge to create output that is very similar to human language.

Generative AI models sometimes invent information. These inventions are called hallucinations and often sound realistic; it is easy to fall for them. Scientists do not yet fully understand why this is happening.

A program that analyses large amounts of data and independently finds solutions without the need to enter a specific code. These types of AI usually generate text, video and other media. ChatGPT is an example of generative AI.

This is a short text excerpt that answers the search query directly on the search results page. They attract a high number of clicks and are suitable for mobile and voice-activated searches.

A form of machine learning in which neural networks are used to find patterns in data.

An AI-generated image, video or audio clip that mimics a person’s image or voice and can be used to spread false information about that person.

GPTs are AI language models developed by the company OpenAI and are based on natural language processing. GPT stands for generative pretrained transformer, which is important because it tells us that this model has the ability to generate or produce output and has been trained on existing data. ChatGPT is an AI chatbot that responds to requests and produces human-like communication in text format.

An AI application that is trained to respond to prompts with conversational responses by text, voice or graphics, usually online. Chatbots simulate a natural conversation and perform certain tasks.

Software that can understand human speech and reproduce it as text or respond to it like a command. ASR can create subtitles in real time using AI-driven technology called natural language processing but is not always accurate yet.

This is the bias that AI programs pick up through the data they are trained with. Racial, gender and cultural biases appear in the data that machines learn from because biased humans create the data. Algorithmic bias can lead to harmful content being produced and is one of the challenges of implementing AI.

A set of instructions that a computer needs to complete a task or solve a problem.