What it is

A foundation model is a massive AI model trained on enormous amounts of data that serves as a starting point for many different tasks. Think of it as a generalist that can be specialised later. GPT-4, Claude, and Gemini are all foundation models. They weren't built to do one specific thing. They were built to be capable of doing lots of things, and then they get fine-tuned or prompted for particular uses. Building one costs hundreds of millions of pounds and requires obscene amounts of computing power.

Why it matters for your job

Foundation models are why AI suddenly seems to be everywhere at once. One model can power a writing assistant, a code generator, a customer service bot, and a data analyst. Previously, each of those would need its own specialised system. This "build once, use everywhere" approach is why AI capabilities are spreading across industries so quickly. It's not that loads of different AIs were built for your industry. It's that one very capable model got pointed in your direction.

What to do about it

Understand that most AI tools you encounter at work are built on top of a handful of foundation models from a few companies. This means their strengths and weaknesses are broadly similar. Learning to work well with one teaches you most of what you need to work with the others.