How we use artificial intelligence

Artificial intelligence (AI) is a broad term covering a range of technologies which have been developed over many decades. More recently, it has come to include tools powered by large language models, such as ChatGPT, which is based on the GPT series of models.

Some developments, from older technologies like speech-to-text to newer ones such as GPT integration in Be My Eyes, have been important assistive tools for some disabled people.

Large language models

We avoid large language models in our work.

We never deliver work which has been created, in whole or in part, by large language models. When we write code, assessment reports, emails, or anything else which we expect anyone to read, it has been written by a human and is designed for human consumption.

The only times our work has involved large language models have been when we critique their application to specific use cases on behalf of clients.

Other uses

The term AI is a very broad, catch-all term which has applied to many technologies since it was coined in the 1950s. Some of these technologies have become standard parts of computing and business. These include:

• Spell-check and autocomplete
• Fraud detection on payment cards
• Neural voices for text-to-speech
• Transcription and dictation

These are often integrated into things we use. We do use tools like these, but never rely on them fully. This means that, if we deliver, for example, a video with a transcript, the transcript may have had a machine-generated first pass, but it is always read and edited by a human before delivery.