It’s highly likely that if someone is writing or talking about artificial intelligence, whether it’s a marketing manager, SEO specialist or a computer scientist, that what they’re discussing, debating, advising or focusing on is generative AI, a subset of artificial intelligence. This first wave, if you can call it that, represents a watershed moment across the board. Here’s what you need to know.
Generative AI explained
Generative AI is a type of artificial intelligence that is capable of generating – creating, producing, illustrating, developing and so on – new content from scratch. This can be text-based, audio-based and image-based (including videos).
The content is made entirely by machine learning models. Discounting prompts and programming, as well as editing, reviewing and providing feedback, humans aren’t involved in the primary creation process at all. It’s 100% AI-generated.
The content that the AI model creates is, in essence, original. You can think of it this way: you tell the AI model what you want or need and it responds by carrying out, in effect, a substantial amount of research in order to deliver as accurate an output as it can – and extremely quickly, too.
How generative AI works
The reference to research above is just a nice way of thinking about what AI models like GPT-4 are programmed to do. How generative AI actually works in principle is a little different.
The short of it is that the model has been fed an enormous amount of data, which includes text, images, sounds and videos. It is trained to identify patterns, features and characteristics – and how, importantly, everything relates to one another (directly or indirectly. It’s basically making sense of the world from the perspective of humans.
Now that it knows “how things work”, it can produce its own original version when asked to do so. So, if you want it to write a specific blog, such as how to write content for SEO – check out our human-version here – it’ll understand how to structure an article like this and what kind of information to include.
Or, if you want it to create an unusual image, like a t-rex and a robot eating pizza and drinking wine in the style of a Jack Vettriano painting, it will generate an image that is close, as least in its mind, to that description (which is based on what it understands about individual objects and the relationships that exists between them, real or not). It’s impressive stuff.
The quality of generative AI
There’s no doubting how good generative AI is. It’s impressive what it can do and how fast it can do it. What’s a little trickier to answer is how good AI is at creating content – what looks and reads good, for example, isn’t necessarily good (from the perspective of someone who is able to review the output impartially and definitively).
There are two important factors determining the quality of AI-generated content – the prompts and the experience and expertise of the individual either instructing the AI or evaluating the output.
If your prompts aren’t that useful, if you don’t really know what it is you want, and if you’re not capable of objectively and comprehensively scrutinising what has been generated, there is a very strong possibility that the initial output will either be basic in scope, full of mistakes or in need of some work to bring it up to scratch.
What that means is that it’s currently not enough to just accept AI generated content as it comes, especially if you’re relying on tools like Bard or ChatGPT. You need to work with what these tools create and either improve it with, for example, edits, or use it as a starting point to build on. And that’s an important distinction – it’s the human touch that makes AI-generated content better.
The benefits of generative AI
While we’ve already answered this question in a little more depth in a dedicated blog, we’ll summarise some of the main benefits for convenience below.
To start with, speed and scale are two of the best things about generative AI – you can create an incredible amount of content in an extraordinarily short amount of time. And that means you can hit deadlines without fail, get ahead and produce more ideas and content than ever before.
Another benefit is the way generative AI frees up your time for other, more complex and productive activities that you’ve typically been unable to get around to. When used effectively, generative AI solutions effortlessly help everything become more efficient, from brainstorming ideas and writing outlines to creating social media posts and scroll-stopping ads.
Finally, generative AI helps deliver faster results. For example, you can test a lot more a lot quicker, get insights sooner than you normally would, and respond to opportunities more promptly and flexibly. That speed means you can work out problems more rapidly and double down on activities that are working at speed.
Examples of generative AI solutions
Text-based generative AI
You name it, ChatGPT can create it, from inspirational blogs and landing pages to ebooks. Poems? Check. Short stories? Check. Essays? Check. Technical content? Check. It can even write code. Discover more about it here.
Similar to ChatGPT, Google Bard is capable of generating a whole host of text-based content, whether you just want to speed up the ideation process or you urgently need to draft a press release. Think of it as a collaborator.
Our AI-powered content solution has been designed to help marketplaces and ecommerce brands generate highly effective and personalised content at speed and at scale. Think optimised products and SEO-rich landing pages. Here’s an overview.
Image-based generative AI
You can think of DALL-E as the image-equivalent of ChatGPT (no surprise, really, as it has been created by OpenAI). It basically works the same – you provide it with a prompt outlining the kind of image you want and the machine does its best to make it happen in a near instant.
One of the leaders in this generative AI visual space, Midjourney can also quickly and effectively generate highly complex imagery based on the prompts you give it. It seems to be very good at creating highly photorealistic images.
Although not as familiar a name as DALL-E and Midjourney, Stable Diffusion is one to watch out for. Again, the principles are the same – you enter a prompt and the AI swiftly responds with as accurate an output as it can.
Video- and audio-based generative AI
Sure, it’s easy creating and sharing videos – but doing it well and professionally, well, unless you’re a proper pro, it will cost you a few bob. Synthesia is one of the brands looking to change that with its stock AI avatars. It’s promising but could do with some more work.
Whether you need a tune for an ad or you want a certain type of music to feature in your about us video, Boomy, with the power of AI, lets you create all types of sounds. It’s possible you’ve heard a Boomy tune, as it claims its users have created over 14 million songs.
Murf.ai’s speciality is in the text to speech space – think voiceovers and audio for podcasts, explainer videos, social media ads and more. You provide the text and it provides the audio in a matter of minutes, which you can tweak as you see fit.
From helping businesses that have always struggled to justify marketing spend to now do more for less to providing organisations with new new resource without the extra costs that might associated with that – provided by, for example, the streamlining of production processes – the opportunities that generative AI presents is extraordinary.
But it’s not just the technology that makes it remarkable. What really brings generative AI to life – and, ultimately, what makes it extremely useful and effective – is the insight, talent and skill of the individuals who are able to harness its power to their advantage. And that’s how you get to be in a totally different league – with people.
Get ahead of your competition with AI-generated content. To find out what that might look like for you, drop us a note and we’ll be in touch.