AI glossary · Fundamentals
AGI
In one sentence
AGI (artificial general intelligence) is a hypothetical AI able to learn and perform almost any intellectual task a person can, with the same flexibility. There is no universally agreed definition and no system widely accepted as AGI.
Also known as: artificial general intelligence · strong AI · general AI · human-level AI
What is AGI?
All the artificial intelligence you use today is, strictly speaking, narrow. One system transcribes, another translates, another chats. Even large language models, which seem to know a bit about everything, have clear limits: they can be confidently wrong, they don’t remember what you taught them yesterday unless it’s stored and fed back, and they struggle to plan long tasks unaided.
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 pricingAGI would be different in kind: a system that picks up whatever it needs for any new task and handles it as fluently as a person. It’s the stated goal of several AI labs, and a hotly debated term.
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’s defined (and why that’s hard)
There’s no single definition. The most quoted ones revolve around three ideas:
- Generality. Doing well across a very wide range of tasks, not just the ones seen in training.
- Learning. Picking up new skills from few examples, and doing so continuously.
- Autonomy. Setting its own intermediate steps and finishing long pieces of work without constant supervision.
Then there’s measuring it. The exams used to compare models keep getting beaten, and every time a system passes one, people argue over whether that shows generality or just practice at that kind of test. So claims that AGI is “nearly here” depend heavily on what you mean by it.
Example
An architecture practice in Bath already uses several AI tools: one transcribes client meetings, another summarises planning policy, a third drafts design and access statements. Each does its part; the team stitches the pieces together, checks them and makes the calls.
An AGI, if it existed, could in theory run the whole process: understand the brief, check the relevant planning rules, propose a design and revise it after the meeting. That isn’t how things stand today, and no serious vendor should promise you otherwise.
What it means for your business
AGI makes for an interesting debate, but it’s no reason to put decisions on hold. What’s useful today is narrow AI applied well: an AI agent with a clear goal, specific tools and rules you control.
That’s UTIO’s approach. The WhatsApp agent and voice agent carry out defined tasks, such as answering, booking or qualifying leads, and pass anything outside their remit to a person. The tasks already running in production are set out under verticals, with no promises of general intelligence.
Frequently asked questions
Does AGI exist yet?
There's no consensus that it does. Current models do many tasks very well but still fail at others a person finds easy, and they don't keep learning on the job the way an employee does. Some companies say they're getting close; no accepted test confirms it.
What's the difference between AI, AGI and superintelligence?
Today's AI is narrow: it handles specific tasks. AGI would match a person at almost any intellectual task. Superintelligence, also hypothetical, would far exceed people in nearly every field.
When will AGI arrive?
Nobody knows. Estimates from researchers and companies range from a few years to several decades, and depend heavily on how each defines AGI. Treat any specific date with caution.
Related terms
- Artificial intelligence
Artificial intelligence (AI) is the branch of computing that builds systems able to do tasks we associate with human intelligence, such as understanding language, recognising images or making decisions, by learning patterns from data.
- 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.
- Reasoning model
A reasoning model is a language model trained to produce a chain of intermediate steps before giving its final answer. It does better on logic, maths and planning problems, at the cost of slower replies and more tokens used.
- Agentic AI
Agentic AI is AI that pursues a goal instead of just answering: it plans steps, uses tools, checks the results and decides what to do next with a degree of autonomy, within rules set by a person.
- Benchmark
An AI benchmark is a standardised test, with a set of questions or tasks and a fixed scoring method, used to measure and compare how different models perform at a specific skill such as reasoning, coding or answering questions.
Updated on