AI glossary · Fundamentals
AI model
In one sentence
An AI model is the program you get by training an algorithm on data: a set of learned parameters that, given a new input, produces an output such as a prediction, a label, a piece of text or an image.
Also known as: artificial intelligence model · machine learning model · ML model · trained model
What is an AI model?
Three things often get lumped together. The algorithm is the learning method. The model is what you get by running that method on data: a file holding millions or billions of numbers, its parameters, which store what it learned. The application is what you actually see: a chat window, a translator, an agent that answers the phone.
Put it to work
AI agents that look after your customers
From €49/mo + VAT
Voice, WhatsApp and email with the phone line included: agent, numbers and CRM in one fee.
See plans and pricingA model doesn’t store its training data like a database. It stores patterns. That’s why it can handle cases it has never seen, and also why it can be confidently wrong when it lacks information.
Related to this term
UTIO agents
- Available 24/7 in 26 languages
- Set it up yourself for €0: built from your website
- 14-day free trial, no card needed
From €49/mo + VAT
See plans and pricingHow it works
A model’s life has two stages:
- Training. It’s shown lots of examples and its parameters are adjusted until it gets them right. This is expensive and done rarely. See model training.
- Inference. The trained model takes a new input and computes the output, every time someone asks it something. See inference.
Some models are small and specialised, doing one job quickly and well. Others are large and general-purpose, like the language models behind generative AI. Choosing between them is a trade-off between quality, speed and cost.
Example
An estate agent in Manchester wants three things: transcripts of buyer enquiry calls, short summaries of each one, and the key details pulled out of scanned documents that sellers send in. That isn’t one model, it’s three: a speech-to-text model, a language model to summarise and an OCR model for the documents.
The agency trains none of them. It uses them off the shelf and adapts them with clear instructions: which details to extract, in what format and where to store them.
AI models at UTIO
With UTIO’s agents you don’t pick or pay for models separately. For the voice agent, the minute includes the model, the voice, transcription and telephony. You configure what matters to the business: what the agent knows, what it may do and when it hands over to a person.
If you build your own tools, UTIO’s AI API gives you transcription, text to speech, text, image, OCR and embeddings with a single key. For text, you choose the model tier (basic, standard or superior) depending on how you want to balance quality against cost for each task.
Frequently asked questions
What types of AI models are there?
By task: classification models (spam or not?), regression models (how much will I sell?), language models (understand and generate text), speech models (transcribe or speak), vision models (interpret images) and multimodal models, which combine several kinds of data.
Is ChatGPT an AI model?
ChatGPT is an application. Under the hood it runs one or more of OpenAI's GPT language models. The distinction is useful: the same app can switch models, and the same model can sit inside many different apps.
Can I use an AI model without training it myself?
Yes, and that's what most businesses do. They use models already trained by a provider, through an API or inside a product, and adapt them with instructions and their own documents, without retraining anything.
Related terms
- Algorithm
An algorithm is a finite sequence of well-defined steps for solving a problem or completing a task. In AI, the algorithm is the method that learns from data; the model is what that method produces.
- Model training
Model training is the process in which an AI model learns from a dataset: it's shown examples, its errors are measured and its parameters are adjusted, over and over, until its outputs are good.
- Model parameters
A model's parameters are its internal numeric values, mainly the weights of a neural network, which are adjusted during training and store what it has learned. Their count is used as a rough measure of model size.
- Inference
Inference is when an already-trained AI model is put to use: it receives a new input, such as a question, an audio clip or an image, and computes a response from what it learned, without changing its parameters.
- LLM
An LLM (large language model) is an AI model trained on vast amounts of text to predict the next word. That single skill lets it draft, summarise, translate, classify and hold a conversation.
Updated on