AI Powered Chat & Search

Search is probably the most obvious application and the one which ChatGPT has brought to prominence. ChatGPT has certainly ruffled some feathers over at Google and they’ve rushed (and botched) the announcement of Bard AI to counter the growing affinity ChatGPT has enjoyed.

ChatGPT on the surface may seem like just a simple conversational interface placed on top of a large amount of information (ironically gleamed from Google) but the power of ChatGPT is, in fact in what they call the language learning model. As more people interact with the bot it is analysing these conversations and learning to understand how humans converse. If this model is successful, over time the accuracy and delivery of its response to all questions will improve and eventually it could become an all-knowing entity with a personality (kind of) and a love for chatting. This is powerful and potentially a google killer.

For Asset Management marketeers this type of tool could certainly be useful to generate long form content quickly (as a research aid) and be sure that it’s relevant. However are the ChatGPT solutions ready to be relied on to take over this important role?

From my own experience, ChatGPT and its peers have still got a way to go before they approach a viable alternative to search engines. Here’s two examples to illustrate the broad scope of success using the tool. Our developers have successfully been using the tool to locate APIs for a variety of services we’re incorporating into our platform builds and doing so with remarkable first-time accuracy. However, on the opposite end of the scale, when I asked ChatGPT to tell me simply who won the last Rugby World Cup it told me New Zealand won the World Cup in Japan in 2019… the wrong answer.

The takeaway here for me is that we’re not there yet in being able to trust a tool like ChatGPT to provide accurate and relevant content for us or even just 100% assured answers to simple queries during research.

AI content writing

AI content writing platforms do exist and have done for a while now. However they’re based on older AI implementations which focus on concepts of relevancy, sentiment, buzz etc to extract and rewrite content for use online or elsewhere.

It’s highly debatable that this content will provide a level of insight into topics which will meet the objective of positioning our asset management clients as thought leaders in their various areas of focus. Further to this I’d imagine compliance officers having a field day (or sleepless nights) at the concept of content a firm publishes written essentially by a machine.

Creative assets and visual design

Developing creative messaging platforms and extending them into impactful visual assets like advertising banners or digital brochures is what our agency excels at and this requires a level of creativity that we draw from our talent team of people. Is creativity something that a machine / AI can replicate? For me, creativity is partly the ability to think beyond the usual and even if starting with an existing concept to push it further than it’s been before to generate the most powerful outcomes. This creativity is not just taking in existing ideas and concepts and trying to push them together which is what an AI creative machine brain would surely be doing.

The only examples I’ve seen to date of AI creativity were algorithms being given a concept and outputting a matching image. It was novel and at times you might have thought creative. Yet it clearly was not what we’d describe as “good” creative. The outcome appeared shiny and overworked and lost it’s appeal soon after first glance.

I still believe creative thinking and execution is therefore nowhere close to being a capability that can be replicated by machines.

When it comes to the creation of complete digital ad sets are we not only at a point where a machine can take one of a large set of stock images, recolour to brand and place some text from a prewritten library and unashamedly call it AI generated advertising?

Once again, I think compliance may have a view on whether this type of solution would get a knock on the door.

So, is the AI powered marketing future imminent?

I’ve not even touched on the bigger concerns with AI regarding ethics, safety and ownership and already you can see there is need to think twice before being swept away on the AI wave.

Yes, some of the tools / applications discussed here would certainly have a dramatic effect on the timing and cost of creating content, however there is no degree of confidence in the quality of the outcomes. That means there is a question about the effectiveness of existing tools to supply effective outcomes. The bottom line is that a business like ours is built on the strength of our creativity and to date I’ve not seen Artificial Intelligence demonstrate creativity to a standard required in a business like ours.

Perhaps the next generation of AI tools will be able to replicate the creative brain but then again how do we replicate what we know so little about?  I am not convinced we’re staring down the barrel of the AI future in marketing but remain open to embracing it if the evidence says differently.

Luke Delport
Director of Digital Development

Credits

1. https://www.techtarget.com/searchenterpriseai/definition/AI-Artificial-Intelligence