From the course: Generative AI Skills for Creative Content: Opportunities, Issues, and Ethics

Limitations in creativity and capabilities

From the course: Generative AI Skills for Creative Content: Opportunities, Issues, and Ethics

Limitations in creativity and capabilities

“

- While generative AI may seem like it's uncovering vast new frontiers of creativity, it's also creatively limited. That's largely because the models can't actually produce brand new ideas. They make associations by using past data to inform future creative work. This can prevent the creation of results that are truly original or innovative. Along those lines, AI models are limited by their own algorithms. We know the algorithms use probability distributions to create content that is similar to the training data. Asking the AI model to think outside the box and consider something that is wholly different from the training data often creates unexpected, undesired and yes, uncreative results. There can also be significant limitations around prompting. AI models can become easily confused if your prompt is too complex or too niche. It can also often take many attempts to tweak your prompt language to get the AI model to understand exactly what you want it to do. This may lead us to settle for something that's close enough rather than getting it right. Contrast this with a human's ability to hone in on perfection and there's no contest to which is more creative. Another limitation is around recency. As we talked about, many AI models that don't tap into the power of the internet aren't able to access recent data. This can be very limiting if you want to incorporate ultra timely ideas or concepts into your creative work. What's worse is this limitation isn't always apparent in the AI model's results, so the lack of recent data may produce creatively inaccurate content. Finally, generative AI models aren't really able to describe their creative process. This means we can't truly understand what went into creating the result, which makes it difficult to break down individual parts of the process or to replicate the process to generate new content. So yes, as a rule, generative AI is no match for the power of human imagination. But when used as an ideation and research tool rather than as a tool to create final works, the supercharged possibilities of AI and human creativity can be a truly awesome combination.

Contents