Why Human in the Loop Matters for AI in Creativity

Generative AI is a powerful technology that is changing the way we are working.  It is so powerful, many are asking about the likelihood of the technology replacing humans in the workplace, especially in the creative sector. That is why Human in the Loop is an important process to understand when it comes to adopting AI in your business. 

Human in the Loop, or HITL, is the process of human intervention in workflow when using AI. I am going to focus on generative AI as that is the technology that will potentially have the biggest impact in the creative sector.

I get it - I’ve seen the posts on social media and I’ve heard from customers directly what generative AI means for them. It is mostly a negative sentiment around the technology. The concerns are real and understandable. If someone can just write a prompt that generates a block of creative text, or draw an image in seconds, who needs a copywriter, or a graphic designer?

Emotions and numbers

Creativity is invention from the imagination of the human mind. Generative AI is an impressive but hugely complex technology around numbers and probability. The former is influenced by emotions, the latter is influenced by calculators and programs. When we look at a painting, we see the colours and the imagery, which in turn evoke emotions inside of us. Emotions that an artist feels when they create the piece of work. When we watch a film, we connect with the characters.

While AI can recognise patterns, it lacks any form of empathy. It is able to replicate human output, but unable to replicate what makes us human. If we’re to explore the philosophical nature deeper, AI is unable to replicate the human soul. 

Input and Output

When we prompt AI, we ask it to perform a task. If we ask it do something basic for us, it will return something basic. Without context, it’s a bit useless. For example - I could ask it to write a blog article about donuts, and it will do exactly that. But what if I wanted it to write from a different perspective, or about a specific type of donut, in over 5000 words? Again, it will do the task. But maybe I’m not being specific enough, or the output is undesirable? It doesn’t fit your style of writing, but you can’t quite explain your style of writing. 

This is why Human in the Loop is absolutely vital when it comes to working with generative AI. A Human -in the Loop approach ensures AI output aligns closely with your needs. I’m also not suggesting that you can hire any human to create your prompts to get your perfect output, I mean a human who knows their craft, or at least some fundamental understanding of what they are creating. 

At your service

This is why I am positioning generative AI as a production assistant, rather than a replacement for the professional. It is your second pair of eyes and ears over a piece of work that you are creating. It is your tool that ensures that you are creating your best work.

For when an individual is creating their best work, they are maximising their value. If an individual is working for an organisation, that organisation also benefits. Some of the benefits include:

• Visualise quicker - Translating ideas and putting them on paper can take time. Generative AI can help creatives visualise their ideas more efficiently. For example, imagine a creative wondering what a picture of a horse inside a car might look like - generative AI can bring that radical idea to life instantly, helping determine whether it works visually or needs refining.
• Be inspired - During the visualisation process, generative AI can inspire new ideas. For instance, if you generate an image of a horse in a car but the AI places the horse in a bus instead, the result might actually look better than your original, wacky idea. This unexpected creativity can lead to fresh perspectives and concepts.
• Prototype - Providing prototypes or proof of concepts during project proposals can be time-consuming and is often unpaid. Generative AI can streamline this process. For example, you could generate several images of a horse in a car but also include one of a horse in a bus. If the client prefers the bus concept, you’ve saved time and effort while delivering a valuable option, freeing yourself to focus on paid production work.
• Summarise meetings - Briefing meetings can be overwhelming, with clients sharing a lot of ideas that can be hard to translate into actionable steps. Generative AI can summarise key points from these meetings. Tools like OtterAI and CoPilot can even join meetings in real time, capturing and summarising discussions so you can focus on the creative work rather than note-taking.

AI isn’t perfect

AI isn’t perfect, and it never will be. Which is another reason why, when AI is adopted in the workplace, Human in the Loop is essential. The technology will always need guidance, and will never be in the position of being able to operate by itself. Some of the limitations include:

• AI can, and will, get it wrong - there is no doubt about it, AI can, and will, get it wrong. While it is an incredibly powerful tool, you are also advised to do your own research.
• AI can hallucinate - AI hallucinations are when AI really does it wrong. Call it a bug. It can either end up giving you information that is simply not correct, or its output is completely random and completely unrelated to the task you gave it to perform.
• AI doesn't know everything - AI is limited to what it knows. While most models now will search the internet when it doesn't have the answer, that doesn't necessarily mean it will look in the right places.
• AI can only replicate - AI is built on probability and programming. It knows what 'things look or sound like', but in many cases, it won't know the theory behind the creations.

Conclusion

Human in the Loop is a vital model to follow when adopting AI in the workplace, especially if you are working in the creative sector. We shouldn't fear generative AI - we should harness the power of this new technology to inspire us to do our best work, but also to make us more productive as human beings. Human in the Loop helps us achieve those goals, but also re-assures us that we're constantly involved in the AI process.