“Generative AI”, you might have heard or seen this word recently and perhaps you might be curious about what generative AI meaning really is. As you notice, the word tends to be associated with ChatGPT, or if you are from creative fields or industries, perhaps you may have heard this term from Adobe’s new feature for Adobe Photoshop, the Generative Fill.
In a nutshell, generative AI is an artificial intelligence that can create various outputs ranging from text, image, and even audio samples from various pieces of information that are being “fed” to its algorithm. This article will dive further into generative AI definition and examples.
What is generative AI?
As previously mentioned, a generative artificial intelligence (AI) is just like what it sounds, an AI that is capable of generating various outputs using tonnes of information that the creator fed to its AI algorithm. That information that has been fed to the algorithm then being used to train and learn by the AI to generate outputs in accordance to the users’ requests.
The step of providing tonnes of information to the AI is important because it is meant to build its “neural network”. The easiest way to imagine this step is let’s say we wanted the AI to recognise what a cat is. We can achieve such a result by providing plenty of images of cats as references to the AI by tagging those pictures accordingly. With enough information, it will eventually be able to distinguish the features of what cats usually have.
In short, it is kind of loosely similar to how our brain works in the sense that we humans also need to learn and be given various examples to finally understand certain subjects.
Now that we know the concept of generative AI and how it can recognise and process users’ requests, we can then see what the AI can output and how many generative AI companies that are currently developing this technology.
Examples of generative AI
According to Sequoia generative AI, there are several models of this type of AI; text, code generation, images, speech synthesis, video and 3D models as well as several other potentials in the future.
The ChatGPT developed by OpenAI for example, is one of the now plentiful generative AIs that are capable of fulfilling text and code generation models as the basis of this AI is to be a “chatbot”. We have previously made 2 articles related to this that you can read in 4 Free AI Code Generators to Assist Your Coding Needs and 5 Recommended AI Essay Generators to Help Your Writing which covers several AIs that we can use to increase our productivity in the field of writing and coding.
DALL-E, which is also developed by the same developer who made ChatGPT, OpenAI, is one of the examples of generative AI that are engineered to generate visual images through natural language processing (NLP). There are also couples of generative AI that are bundled with other functions such as Novel AI which aside from being capable of text writing, can also generate images.
In the creative field, there’s also Generative Fill from Adobe’s Photoshop which we previously mentioned in the intro. Aside from its ability to generate visuals through text, Adobe also combines the strong editing point of Photoshop which enables users to fix or adjust the generated visuals to suit their needs manually with ease.
Generative AI videos have also started to become a bit of a phenomenon recently. Runway, an AI startup focusing on the video aspect of generative AI, was the first company who let users try out its video-to-video generative AI model through a normal smartphoneー In fact the app itself is still available for free in the app store.
The end result of its first public release, dubbed as GEN-1, works kind of similarly to Instagram or Snapchat filters, however, in its recent development under the name GEN-2, it introduces the text-to-video method which lets users to put NLP prompt and the AI will do its best to interpret them to video output.
Generate videos with nothing but words. If you can say it, now you can see it.
Introducing, Text to Video. With Gen-2.
— Runway (@runwayml) March 20, 2023
As for speech synthesisers, there are two mainstream AIs made by Uberduck and Murf. While both mostly function the same, Murif is tuned to focus on voice-over aspects while Uberduck is meant for music creation. What is interesting with Uberduck however, it lets users make their own voice bank to experiment around.
Currently, it is safe to say that we are currently in the midst of a technological leap in artificial intelligence. In the generative AI aspect alone, there are already several models that are rapidly being developed with really quick progress in just a matter of months, weeks even.
A few of those models are in the form of text including code, script, and even essay generation, images, speech synthesisers, as well as video and 3D models.
On the definition, generative AI is an artificial intelligence that can create various outputs as mentioned above using the information from AI’s trained neural network.