2/1/2023
But, as with any emerging technology, it comes with its own set of risks, so naturally we wanted to know more. To conduct a thorough investigation into this new frontier, we sent out tech strategist and resident AI specialist Deevid De Meyer to research what exactly creative AI is capable of and what risks come along with its implementation.
A lot of my marketing colleagues were perhaps somewhat rightfully concerned for their jobs when I first presented the possibilities of GPT-3(.5) based language generation. After all, AI-based copywriting tools can produce pages and pages of content with a fraction of the effort and cost. Moreover, at times even the quality of content outperformed what we could come up with ourselves. To illustrate my point, I even had GPT write a limerick on the situation:
There once was a battle between
Human and AI creativity seen
One with emotions
Other with notions
But both can be winners, as a team
Fortunately, after experimenting with these tools, I think that my colleagues have come around to understand a key feature of any machine learning system: they're not always right, and therefore you often need a trained eye to determine the quality of the results.
Using an AI language tool is like having a very enthusiastic junior colleague with an impressive vocabulary and a tendency to blurt out whatever they think is the right thing to say. It can be a great resource, but you still need human judgment to separate the great from the mediocre - and sometimes downright wrong.
At Unmute-You, our biggest client base is tech companies. This means that a big part of our job boils down to translating technical concepts into human language that properly conveys the business value of these concepts. And it's not just about getting the facts right, but also about conveying enthusiasm and energy in a way that will capture the attention of potential customers.
Language generation models are very good at the latter, but struggle when it comes to presenting factual statements. AI-powered copywritings often make bold and convincing statements, but have no issue completely making up facts as they go. So, while these models can be great for generating creative content, you need to constantly exert editorial judgment to ensure accuracy and relevance of the statements made.
While we've primarily discussed language-generating AI tools thus far, the future of creative AI extends beyond text. Image-generating AI models, such as DALL-E2 and Stable Diffusion, have captivated us with their ability to produce images based on human input.
However, language understanding remains crucial in interpreting and refining this input. The integration of text, image, and video generation tools holds immense potential for an omnichannel content creation experience, representing the future of AI-powered marketing.
Creative AI tools are a powerful resource for marketing departments. They can be used to create content faster, cheaper and with less effort. However, it’s important to understand the potential dangers of these systems - such as confidently incorrect statements - and use them responsibly. My advice is to focus on using AI to enhance existing human processes, rather than trying to replace them. This way, you get to reap the benefits of AI automation without running the risk of falling into one of its pitfalls.
So by now you must be thinking: "An article written about language generation, surely the big twist at the end will be that the entire thing has been written by an AI all along?"
Well, to be fair, I tried. But the AI kept falling back into clichés like 'human ingenuity is still unique' and 'AI's will never be able to surpass human creativity'. Perfectly illustrating my point, because if I hadn't known as much about AI, I would have probably been content with cliché – or downright wrong – statements like the above.
- Contact us