AI = Absence of Imagination?

The Messagery's two Davids examine AI’s role in message development

A lot has been written about ’AI‘, often by people with great imaginations but few experiences of the topic. The two Davids at The Messagery have, over many years built, marketed, played with and reported on software that demonstrates some ability that people called artificial intelligence.

Without doubt, the current crop of generative ’AI‘ tools such as ChatGPT, Bard, etc. are exhibiting exponential growth in perceived capability. That’s primarily due to the extraordinarily large amounts of data being ‘fed’ to them, along with considerable amounts of raw computing capability.

But here’s the first limitation: These AI systems are limited to the data set that’s available to them. They can assemble an ‘answer’ from what they ’know‘. They cannot speculate, have limited deduction capabilities and simply can’t deal with anything newer than their available data set.

We interviewed ChatGPT some time ago and asked it (amongst other interesting questions) whether AI can be truly creative, the answer seemed to boil down to the idea that it can “exhibit elements of creativity”. You might enjoy reading the full interview after you’ve finished here.

You’re probably here because of your interest or involvement in messaging so, at a messaging level this means that you can’t expect one of these tools to be able to develop strategic or tactical messages for a product or service that is new or unique. The data isn’t available, it’s in your company’s collective heads. Only you will have the insight, inside knowledge and imagination, to create those messages.

However, you can persuade these tools to give you overviews of any topic you wish to explore. Using ‘live’ data from their users, they may even be able to deduce which topics are ‘hot’, and which are ‘not’. It’s a shaky foundation to use for generating material.

ChatGPT itself points out that it uses “a mixture of licensed data, data created by human trainers, and publicly available data”. It also claims a hands-off relationship with the training material. Open AI has a procedure for copyright infringement claims. To be on the safe side, be informed by AI not be ruled by it. We are already seeing unscrupulous people submitting AI-generated work as their own.

We’ll see how laws respond to challenges such as these, but they are lagging. Perhaps the equivalent of the GDPR ’Right to be forgotten‘ can be applied to training content. As ever, the unscrupulous ‘bad actors’ will ignore any regulations.

Our take on this? AI Chatbots are no substitute for insight and ‘new’ ideas. They won’t replace imagination. That’s not an excuse to reject them, they have a place in assembling ’what is known‘ into relevant, coherent and easily digestible form. The skill lies in asking the right questions.

Hopefully, we did.

Image credit: José Manuel de Laá