The Glossary TechCrunch AI | Technological

Artificial intelligence is a deep and complex world. Scientists who work in this field are often based on terminology and lingo to explain what they are working on. As a result, we often have to use these technical terms in covering the artificial intelligence industry. That is why we thought it would be useful to gather a glossary with definitions of some of the most important words and phrases we use in our articles.

We will regularly update this glossary to add new entries, as researchers will constantly reveal new methods to promote artificial intelligence borders while identifying emerging security risks.

An AI agent refers to a tool that uses AI Technologies to perform a series of duties on your behalf – beyond what could make a more basic AI Chatbot – such as archive costs, booking tickets or a table at a restaurant or even writing and code maintenance. However, as we have explained earlier, there are many animations in this pop -up, so different people can mean different things when referring to an AI agent. The infrastructure is still constructed to achieve the opportunities intended. But the basic concept implies an autonomous system that can derive from multiple AI systems to perform multiple -step tasks.

Given a simple question, a human brain can answer without even thinking about it – things like “which animal is taller between a giraffe and a cat?” But in many cases, you often need a pen and paper to find the right answer because there are intermediate steps. For example, if a farmer has chickens and cows, and together they have 40 heads and 120 feet, you may need to write a simple equation to find the answer (20 chickens and 20 cows).

In an AI context, reasoning chain of thinking for large linguistic models means that the split of a problem into smaller, intermediate steps to improve the quality of the end result. It usually takes more time to get an answer, but the answer is more likely to be correct, especially in a reasonable or coding framework. So -called reasoning models are developed by traditional large linguistic models and are optimized for thinking chain thinking thanks to enhancing learning.

(See: Linguistic model;

A subset of mechanical learning in which AI algorithms are designed with multiple layers structure, artificial nervous network (ANN). This allows them to make more complex correlations than mechanical learning -based systems, such as linear models or decision trees. The structure of deep learning algorithms draws inspiration from the interconnected streets of neurons in the human brain.

Deep learning AIS is able to identify important characteristics in the data itself, instead of demanding human engineers to determine these characteristics. The structure also supports algorithms that can learn from errors and, through a repetition and adjustment process, improve their own outputs. However, deep learning systems require many data points to give good results (millions or more). Also, it usually takes more time to train deep learning against simpler mechanical learning algorithms – so the cost of development tends to be higher.

(See: Nervous system;

This means further training an AI model intended to optimize performance for a more specific task or region than previously was a focus of training-usually with diet in new, specialized data (ie work-oriented).

Many newly established AI companies receive large language models as a starting point to build a commercial product, but are struggling to enhance the utility for a target sector or work by filling in previous training cycles with refinement based on their own specific knowledge and expertise.

(See: Large linguistic model (llm);

Large linguistic models, or LLMS, are the AI ​​models used by popular AI assistants, such as Chatgpt, Claude, Google’s Gemini, Meta’s AI Llama, Microsoft Copilot or Mistral’s Le chat. When chatting with an AI assistant, you interact with a large linguistic model that processes your request directly or with the help of different tools available, such as internet touring or interpreters.

AI assistants and llms can have different names. For example, GPT is Openai’s large linguistic model and chatgpt is the AI ​​assistant product.

LLMS are deep neural networks made of billions of numerical parameters (or weights, see below) who learn the relationships between words and phrases and create a representation of the language, a kind of multidimensional map of words.

These are created by the encoding of patterns found in billions of books, articles and transcripts. When urging an LLM, the model creates the most likely pattern that fits the prompt. He then evaluates the most likely next word after the last on the basis of what was said before. Repeat, repeat and repeat.

(See: Nervous system;

The neuronal network refers to the multi-layer algorithmic structure that supports deep learning-and, more broadly, the entire explosion of AI genetic tools after the appearance of large linguistic models.

Although the idea of ​​being inspired by the dense interconnected streets of the human brain as a design structure for data processing algorithms dates back to the 1940s, was much more recent rise in graphic processing material (GPU) – through the video game industry – which really unlocks the theory. These chips proved to be suitable for training algorithms with many more layers than possible in previous seasons-allowing AI systems based on the neuronal network to achieve much better performance in many areas, either for voice recognition, autonomous navigation or drug navigation.

(See: Large linguistic model (llm);

Weights are the core for AI training as they determine how much importance (or weight) is given to different features (or input variables) to the data used to train the system – thus shaping the output of the AI ​​model.

In another way, weights are numerical parameters that determine what is more important in a data set for the given educational project. They achieve their operation by applying multiplication to entrances. Model training typically begins with random weights, but as the process unfolds, weights are adapted as the model seeks to reach an output that best suits the target.

For example, an AI model for predicting housing prices trained in historical real estate data for a target site could include weights for features such as the number of bedrooms and bathrooms, regardless of whether a property is detached, semi-defined,

Ultimately, the weights that the model is connected to each of these entrances are a reflection of how much they affect the value of a property, based on the given data set.