AI is here to stay – meaning you’re likely to be hearing a lot more about the technology over the coming months and years. So here are 10 AI-related terms you may not yet know, but which it may be worth getting familiar with.
1. AI
To start with, then, AI itself. This is simply the umbrella term covering any Artificial Intelligence system that mimics human knowledge in a particular way (everything from chatbots to the more old-hat translation services).
2. Machine learning
AI and machine learning are sometimes seen as interchangeable but that’s not quite the case. Think of it more that AI is the destination, machine learning is the way to get there.
3. Large language models
To accurately gain ‘knowledge’, AI systems need to process enormous swathes of information. One system (the large language model) uses huge amounts of text not just to learn facts and figures but also the ways in which humans speak, write and understand one another.
4. Multimodal models
The ‘triple threat’ of the AI world, multimodal models can work across different types of data simultaneously (be that images, text or sounds). This multi-pronged approach is used for tools with specific tasks like answering questions about photographs.
5. Generative AI
Those headline-grabbing AI tools that create images or write poetry are examples of generative AI, where a system learns patterns and structures to create something new.
6. Hallucinations
Generative AI tools are impressive, but aren’t yet totally trustworthy. The systems can’t work out fact from fiction so will sometimes give wrong information – aka a ‘hallucination’.
7. Prompts
The inputs you provide into an AI system to steer it towards the outputs you need are referred to as prompts.
8. Copilot
Not just the name of Microsoft’s own AI assistant, a copilot is one that works alongside you to ease workloads, improve productivity or inform decisions. It’s a strand of generative AI designed specifically for the corporate world.
9. Plugins
As with plugins elsewhere, those in the AI world are designed to help systems interact with others – or with the wider world. They can be deployed to fill a gap or provide a solution without requiring the foundations of an AI system to be overhauled.
10. Responsible AI
Finally there’s responsible AI, the move to ensure that artificial intelligence remains safe and fair – which is particularly important given the expectation that AI will be called on to make more critical decisions as the technology becomes more trusted and more commonplace.
