Looking to the Future: Does ChatGPT Have a Role in Marketing?

Over the last two years, advancements in generative AI have been taking the world by storm, just check out some of the headlines:

Beethoven’s unfinished Tenth Symphony completed by artificial intelligence.

Why It's So Hard to Resist Turning Your Selfies Into Lensa AI Art.

How a deepfake Tom Cruise on TikTok turned into a very real AI company.

ChatGPT passes exams from law and business schools.

‍

Download this POV here.

Generative AI utilises machine learning to create new and unbelievably real digital content with minimal human intervention, leaving many questioning the ethical role this tech plays in our everyday lives. In practice, brands and marketers are assessing how these tools will impact their work and the future of their industries.  

We spoke with some of the brilliant minds at Assembly to get their points of view and what they’re excited about when it comes to generative AI.

What can generative AI be used for, other than creating AI art of cats?

Tim Hawes, Associate Strategy Director for Europe: It has a tonne of utility already and its rate of development is astounding. It seems like a new AI for “X” is released or posted every day. It can pretty much answer any question you throw at it (ChatGPT, that is). It’s like a chat-based Wikipedia in that regard. I’ve asked it to explain history, maths and random facts or even write some marketing headlines or code snippets. It can write recipes – which is what really kickstarted my renewed interest in it. I watched a video where a chef explains her preferences; type of meal and the bot outputs a full menu and instructions for a full Indian-inspired thanksgiving dinner.

Other than cats, it (DALL-E) can create some genuinely interesting art, some can come out looking a little uncanny valley-esque though, depending on what you ask it to generate. There are AIs for everything,  theresanaiforthat.com is a database of many of them; I wouldn’t be surprised if an AI curated it. Because of the way that the ChatGPT  is trained, it can recount any set of information it has ingested, merging sources together and creating ‘new’ text, depending on how you prompt it. You can even ask it to ‘speak’ in a particular mode or instruct it to take on a particular role.

So as far as “what else CAN a generative AI do?” with enough quality input I’d be asking “what can’t it do?”

Are there limitations with ChatGPT and similar tech?

Pedro Mona, Global Director Martech & Data: Two of the major factors limiting scale right now are server capacity issues and computing power – ChatGPT was down recently because half of LinkedIn were trying to access it at the same time. There are also organisations using ChatGPT to churn out essays and sell them to university students, granted the work was brilliant, however, the machine didn’t seem to understand the concept of referencing. So, plagiarism and duplicate content could certainly be an issue – which brings us back to my point on training models specific to clients and brands. On the topic of coding, it's certainly got applications in shortening writing time but it’s essentially not much different from stealing code from the slew of libraries out there anyway – that’s what everyone already does.

David Hidasi, Senior Data Scientist: In the role of data science, it can help us to generate and fill holes in data as well as create basic functions and give us shortcuts for coding – which humans can then elaborate on and develop. We’re not there yet as far as relying on it end-to-end, there will always be some requirement for testing and human curation – given the pre-trained nature of the networks.

What does AI mean for marketers and brands?

Kristie Naha-Biswas, Head of Strategy & Planning for Europe: AI is pretty amazing, but frightening at the same time.  The ability to produce content faster and more efficiently than humans may be appealing to brands or procurement as a new cost-efficient evolution in their marketing deployment, but there is one vital human component that this technology still lacks, which is empathy and emotion. Emotion is what makes art, in any form be that a painter or a musician, unique.  Art is a human expression of emotion that cannot be replicated by AI, it is the artists’ personal experience and original thought that elicits an emotional response from their audience – do we love it or hate it. Advertising creative is no different. Emotionally led and real, insight driven ideas are what makes creative distinct so brands can stand out to build that critical mental availability vital to any successful brand formula. I think there is a future role for how we can use AI to drive greater personalisation of a piece of content, or messages from an overarching creative idea, or concept.

Pedro Mona, Global Head of Data and Martech: The way agencies and brands are going to win with generative AI is integrating it into human-assisted workflows – where maybe an analyst could work on one or two things at a time, now maybe its three or five at once! Training and integrating generative AI models into a brand, where it understands the tone – the voice of the brand – will take it much further to “on brand” content than its current public training models. I see big applications for content in this regard.

The most interesting part is testing human versus AI versus human + AI. In my previous experience producing predictive models for media performance, the human-assisted AI campaign won by a landslide – the input quality and the human context for the brand and therefore analysis bore the best results. While I don’t see it replacing search engines entirely (the ability to index up-to-date data isn’t there yet) it has wonderful applications for accessibility – text to speech rendering and explanation of advanced topics can have brilliant utility for the visually impaired, for example. Ultimately, I think it’s going to become the new standard, improve parallel workflows and companies that can utilise it in the context of their own brands. There will always be an element of human intervention in quality control and analysis that I don’t see going away any time soon.

Download the Full POV here.