Generative AI vs. other types of AI.

Artificial intelligence is the simulation of human intelligence by a machine. Despite ancient mythologising and centuries of efforts to create machines that could act like humans, the first true example of machine learning came in 1951 when a checkers programme at the University of Manchester learnt how to play well enough to beat human players. This was the earliest form of game AI.

Today, AI is everywhere, especially in our phones and mobile apps, where it performs common tasks like facial recognition, speech recognition, spam filtering, grammar suggestions and language translation.

There isn’t a single type of AI that does everything. Just as we use different computer programs for word processing, accounting and digital compositing, there are different types of AI that work differently to achieve different goals. The types of AI that you’re likely to encounter on any given day fall into three basic categories.

This type of AI performs specific tasks and doesn’t learn beyond that. Image recognition, natural language processors and image generators are forms of narrow artificial intelligence. Siri, Google Assistant and Alexa are forms of narrow AI. They can understand and respond to voice commands, but they can’t give you accurate medical diagnoses.

As the name suggests, these machines can only react to stimuli, like the movement of a chess piece on virtual chessboard. They don’t build memories that they can retrieve in the future like limited memory AI machines, but they’re useful parts of digital life. Spam filters and recommender systems that note your behaviour and suggest what you should watch next or what you might like to buy are reactive machines.

The next step in AI development after reactive machines, limited memory AI can store knowledge, learn from it and perform new tasks based on that learning. This type of AI is useful for making predictions and it’s used to forecast future trends in everything from finance to the weather. This is the most common type of AI in use today, appearing in chatbots and self-driving cars.

Just as other subsets of AI are built for specific purposes, generative AI includes different types of models built to produce different types of content.

Text generation is perhaps the most talked about form of AI and chatbot AI models have become popular and appear in the news regularly. People use these text generators for everything from proofreading to writing surveys to creating code.

AI generators like Adobe Firefly can turn a text prompt into an image. They have a wide range of uses for artists, content creators and marketers. Trained on hundreds of millions of images and captions, AI image generators give anyone the power to describe what they want and quickly generate a new image based on a text description.

Artists and designers can use this technology to brainstorm new ideas and try new ways of imagining their work without devoting hours to each iteration. Marketers and independent content creators can produce beautiful images quickly even if they aren’t experienced artists.

Just as image generators are trained on huge datasets of images, AI music generators are trained on music and metadata. They can identify patterns in music genre and the work of specific artists and create new music based on those characteristics.

This form of generative AI combines all three forms of video, sound and image generation. Trained on data that includes audio, video and text, these models can help video producers save time editing, adding captions, resizing for different outputs and even generating new video content based on text prompts.