Is it possible to explain how generative AI works?

Explaining technical concepts in language that people can understand – even if they don’t have a tech background – is hard to do. So I was delighted to see this clear, step-by-step explanation in the Financial Times.

This image from the FT article pulls together the different techniques that have come together in recent years to bring us generative AI tools that can answer our questions or create artwork, music, videos and more.

If you think back to the chatbots we encountered in the past, they were frustrating beasts because they only had a small pool of questions and answers to draw from. This mean they often failed to understand what we were asking and couldn’t answer our queries. These still exist – I encountered one today that insisted I wanted help with setting up a new email address when this was nothing like what I had asked for help with.

But even these basic chatbots have had an important role to play in evolving today’s higher forms of conversational AI. The ability to process many possible permutations of words quickly, and retain information about relationships between words, has taken us into this new era.

This article is well worth a read to appreciate how we’ve got to where we are today, as well as to deepen your understanding of how this type of AI works.

You may also like...