From LLMs to hallucinations, here’s a simple guide to common AI terms

A glossary of essential AI terminology, from large language models to algorithmic bias, decodes the jargon shaping public discourse. | Contesto: cronaca

Punti chiave

• From LLMs to hallucinations, here’s a simple guide to common AI terms

Contesto

The rapid mainstreaming of artificial intelligence has unleashed a torrent of specialized terminology, creating a significant barrier to public understanding of a technology reshaping industries and societies. From boardrooms to legislative hearings, terms like "large language model," "hallucination," and "algorithmic bias" are now commonplace, yet their precise meanings often remain opaque. This lexicon gap complicates informed debate on AI's risks, regulations, and societal impact, making a clear guide to its foundational concepts not just useful, but necessary for meaningful participation in the modern technological conversation. At the core of much recent advancement is the large language model, or LLM. These are AI systems trained on vast datasets of text, enabling them to generate human-like writing, translate languages, and answer questions. They power the chatbots and creative tools that have captured global attention. Closely related is "generative AI," a broader category describing models that create new content—text, images, audio, or code—rather than simply analyzing existing data. The output of these systems, however, is not always reliable, leading to the critical concept of "hallucination." This refers to an AI generating plausible-sounding but factually incorrect or nonsensical information, a fundamental flaw with serious implications for their use in research, journalism, or legal advice. The process of creating these systems involves "training," where a model learns patterns from its massive dataset, and "fine-tuning," a subsequent stage where it is refined for specific tasks or to align with certain guidelines. How a model is directed is managed through "prompting," the art of crafting input instructions to elicit the desired output. Underpinning the entire field is "machine learning," a subset of AI where systems learn and improve from data without explicit programming for every task. A more advanced technique within this is "deep learning," which uses complex, layered neural networks inspired by the human brain to identify intricate patterns. Beyond the technical mechanics, a crucial set of terms addresses AI's societal and ethical...

Lettura DEO

Decisione di validazione: publish

Risk score: 0.0

Il testo è stato ricostruito dai dati editoriali disponibili senza aggiungere fatti non presenti nel record sorgente.

Indicatore di affidabilità

Verificata — Alta confidenza. Fonti affidabili confermano la notizia.

Il sistema a semaforo

Ogni articolo su DEO include un indicatore di affidabilità:

• 🟢 Verificata — Alta confidenza. Fonti affidabili confermano la notizia.
• 🟡 In evoluzione — Confidenza moderata. Alcuni dettagli potrebbero ancora cambiare.
• 🔴 Contestata — Bassa confidenza. Fonti in conflitto o incertezze rilevanti.

Questo sistema esiste perché chi legge merita di sapere non solo cosa è successo, ma anche quanto la notizia è solida.

Categoria: cronaca