Why ChatGPT and Dall-E Will Help Video Creators Be More Creative

The past year or so has seen generative AI explode into the public consciousness – first with the release of Dall-E 2 for creating images and more recently with ChatGPT for creating text. Both of these platforms have spawned multiple imitators like Midjourney and Stable Diffusion, all competing to be the go-to place for helping us realize our creative visions.

As with any major innovation, there are cheerleaders for this technology and those who worry about its repercussions. For people in the creative industry, there’s a particular concern about whether their jobs will still exist in the future.

This technology does come with risks and needs to be used with care. But for creatives, it can actually be an incredible, supportive and empowering tool. 

In this article, we’ll understand the technology, look at its risks, and explore its creative potential.

What is generative AI?

To understand generative AI, let’s first look at the artificial intelligence which we’ve known until now – and which we’re all using on a daily basis. The previous generation of AI is able to understand and classify huge amounts of data. It can then take actions based on that data – such as recommending your next song on Spotify, suggesting the best route to your destination to avoid traffic, or locating a wanted fugitive in a crowded place.

Generative AI is different. It also understands and classifies data, but then uses that knowledge to create new data, completely from scratch. In simple terms, if an AI sees millions of images of cats, it learns how pixels come together to form an image of a cat and can mimic this process to create its own image of a cat – which may look entirely real – but in this case the feline is fake.

What are the risks of generative AI?

When your eyes deceive you

The most obvious risk with this technology is deep fakes. Generative AI can theoretically be used to create images, videos, text and voiceovers that never actually existed, but look, feel and sound real. Taken to an extreme, you could create a video of President Biden declaring nuclear war on Russia – and you can imagine the implications. On a less global scale, jilted partners could create fake revenge pornography or kids could create fake media to bully their classmates.

Arguably, this is just accelerating a situation that is already the case. Mankind has been faking news since the earliest written and drawn records. But, even today, creating fake news requires quite a high level of skill. Generative AI technology could mean that anyone could do it. Imagine a world where you literally can no longer believe your eyes. It’s not a pleasant thought.

Built-in bias

Another risk of this technology is that it will reinforce built-in biases in society. The models train off existing images. So, for example, if you ask for an image of a doctor or CEO, you’re most likely to get a visual of a white man. Some countries and their peoples are much more photographed than others, so they lack representation in the data set. Should the AI reflect the world that we want to see (and who gets to decide what world we want to see?) or the world as it is?

Whose rights?

All creatives rely on having their work protected by copyright law – otherwise, they can’t monetize it to make a living. Most of these generative models – whether they’re creating text, music, images or video – are trained off publicly available data. Is it legitimate for commercial companies to train their models off artists’ work that happens to be online – without licensing it for commercial use?

The technology companies argue that it is legitimate – and that arranging the licensing for the billions of pieces of data they train off just isn’t practical. But many artists don’t agree. But this innovation has moved faster than national legal systems – there’s no clear answer in the law. In the next few years, we should expect several test cases going through the law courts. Until the legal position is clear, some argue that using generated media for commercial purposes leaves you, the user, open to being sued.

It still isn’t reliable

While many of the results published in the media from these platforms are truly incredible, they’re still far from foolproof. Image generators often stray into the “uncanny valley”. This term refers to human likenesses that are very close to being realistic but something is just a little bit off – and they evoke a strong negative reaction in people.

Also, finding the best way to write your text prompt to get the results you want can take time. Even then the results may not work for you. Having said that, with the pace of improvement in these technologies, these issues probably won’t exist in a few years’ time.

How to harness this creative revolution

Despite these risks, generative AI offers an incredible opportunity for creatives. This technology is here to stay, and will be an increasingly important tool for anyone in a creative industry moving forward. Getting familiar with it now and experimenting with how best to use it will place you in a strong position to succeed moving forward.

At Artlist, we’re already using generative AI to improve our search, so you’ll be able to find exactly what you want, fast – even if you have typos or are looking for complex concepts. We’re constantly exploring new applications for the technology. Here are some ways you can use generative AI today.

Research faster than ever before

One prediction is that text services like ChatGPT will replace search engines. The results they return are often more helpful than a similar query in Google. Try asking “where is the best location in the state of Alabama to shoot a romantic film scene” or “how do I plan a film shoot” and see the results.

Inspiration at your fingertips

Ever suffered from creative block? Generative AI tools are incredible for breaking down creative barriers. Here are just a few examples of how you can use them for inspiration:

Get ideas by generating images for

• Character names
• Scripts in different styles (funnier, more serious, more informal etc)
• Dialogue based on specific scenarios
• Titles and captions
• Descriptions for catalogs or award submissions
• Social media posts

Get the AI to do the heavy lifting

In some cases, you can get the AI to do almost all the work for you. Particularly in the case of a text generator where the technology is more polished than with the synthetic images.

For example, get it to generate an email that will promote your video to agents, create the outline of a blog post about your production process or write a legal contract for your suppliers. This won’t get you all the way – you’ll need to edit the email, fill in the structure of the blog post and get a contract reviewed by a lawyer – but it will save you lots of time and effort.

Generative AI is here to stay

In conclusion, this technology is not going anywhere – and if anything, is only going to get more powerful. Harnessed properly, it can really expand your creative horizons and help you work faster, smarter and produce even better video content.

Make sure you don’t get left behind, start experimenting with generative AI today to position yourself at the center of creativity in the future.