Glossary
Tight definitions of the terms that matter — each one connected to where it's actually used.
When a model states something false as if it were fact — a structural feature of next-word prediction, not a bug that’s been fixed.
Term LTV (loan-to-value)The mortgage as a percentage of the property’s value — the number that sets your deposit and often your rate.
Term Token (AI)The chunk of text a language model actually reads and writes — usually a word-piece, not a whole word — and the unit models are priced in.
Term Central bankThe institution behind a currency: it sets policy rates, targets price stability, and banks the banks.
Term Context windowHow much text a model can hold in mind at once — prompt plus reply — measured in tokens; run past it and the earliest text falls away.
Term APR (annual percentage rate)The yearly cost of borrowing including standard charges, used to compare loans and credit cards.
Term APRC (annual percentage rate of charge)The all-in yearly cost of a mortgage including fees, designed so offers can be compared like-for-like.
Term CPI (consumer price index)The basket-of-goods index used to measure inflation — what goes in it and what a CPI figure actually claims.
Term Policy rateThe interest rate a central bank sets on purpose — the lever every other rate in the economy takes its cue from.
Term Basis pointOne hundredth of a percentage point — the unit finance uses when percentage-point moves are too coarse.
Term Fine-tuningTaking a general model and training it further on narrower data to specialise it — cheaper than training from scratch.
Term Gross and netGross is the amount with everything included; net is what remains after deductions — which is which depends on context.
Term AER (annual equivalent rate)What a savings rate really pays over a year once compounding frequency is taken into account.
Term Percentage pointThe unit for the gap between two percentages — the difference between 4% and 6% is two percentage points, not 2%.
Term Purchasing powerWhat money can actually buy — the quantity inflation erodes even while the number in the account grows.
Term Training dataThe text (and other content) a model learns from — its source of both capability and bias, and why a "knowledge cutoff" exists.
Term Equity (property)The slice of your home you actually own: its value minus what is still owed on the mortgage.
Term Parameters (AI)The adjustable numbers a model learns during training — often billions of them — that encode everything it "knows".
Term UTC (Coordinated Universal Time)The world’s reference clock — the time standard every time zone is defined as an offset from.
Term Zero-rated (VAT)Goods taxed at a VAT rate of nothing — which is not the same as being exempt, and the difference matters to businesses.
Term DeflationPrices falling across the economy — rarer than inflation, and more feared by the people who set interest rates.
Term Gregorian calendarThe civil calendar most of the world runs on, introduced in 1582 to stop the seasons drifting through the year.
Term Inference (AI)The act of running a trained model to get an answer — what happens every time you send a prompt, and what you pay for per use.
Term AmortisationPaying a loan down with instalments that cover interest first and principal with the rest, on a fixed schedule.
Term ISO 8601The international standard for writing dates and times — year first, so dates sort correctly as text.
Term Real returnAn investment’s return after inflation — the growth in what your money can actually buy.
Term Compounding frequencyHow often earned interest is added to the balance — yearly, monthly or daily — so it can start earning interest itself.
Term Future valueWhat an amount of money will be worth at a set date in the future, given a rate of growth.
Term PrincipalThe original amount of money saved, invested or borrowed, before any interest is added.
Term