Stay Informed
Sign up here for the latest articles
By Jon Williams
Despite (or maybe because of) the gargantuan amount of commentary out there about AI, the prevailing view for most still seems to be based around apprehension, misinformation and misconceptions.
TRAINING IS KEY
AI shouldn’t be seen as either a quick fix solution or threat to jobs, but as a dedicated research assistant. ‘It’s how you work with and train your LLMs that’ll make the most difference and drive the most value for your organisation.’
Despite (or maybe because of) the gargantuan amount of commentary out there about AI, the prevailing view for most still seems to be based around apprehension, misinformation and misconceptions.
Especially from a value perspective. From thinking it’s plug-and-play with minimal setup, that it can understand nuanced audiences without training or that’s going to steal your job/take over the world (to be fair, the jury is out on the latter) AI can still be a scary and confusing space and no agency, as far as I can see anyhow, has found a truly innovative, interesting or effective way to use it properly.
For me. It helps to think of GenAI as the politest, most hard-working research assistant you’ve ever found. They won’t always get it right. Some of their answers could be terrible, but that’s great.
That’s why your marketing teams won’t be replaced any time soon. It’s how you work with and train your LLMs that’ll make the most difference and drive the most value for your organisation.
And what myself and some really clever people (such as Zach Duenow, former Global Director of Brand Building, Haleon and now founder of MUZE and Tim Sparke, Co-founder at Congregation Partners) have done is look at the key part of the marketing team’s role and generate five simple ways they can use AI to make that process more creative, more effective and more efficient.
In short, more valuable.
One
You have to teach it. Upload your domain expertise – but don’t just drag and drop. Teach it to see the world through your eyes. Give it your perspective and your knowledge of the industry as well as what it finds online.
You can do this by talking to it through your phone. You have way more insight than it can get online, so teach it, talk to it.
Help it learn and to understand not just your knowledge but the knowledge it is picking up from the world. Then, get your team to do the same. It will become an extension of your marketing team.
Two
Then, you have to learn to dialogue with your AI all the time. I know this flies in the face of conventional prompt engineering (and much of human nature), but that is a red herring in this sense.
But you need to talk to it constantly. It thrives on knowledge and you have loads of that. It also thrives on interaction and learning.
Three
You have to try different perspectives. Remember, you are not talking about one artificial intelligence, you are talking about numerous intelligences.
Your LLM can take on a whole plethora of personalities and mindsets, summoning a vast array of thinking models and decision-making avatars, helping you think beyond your own cognitive limitations.
A quick test: “We’re struggling to understand what really drives purchase of Kleenex. Some of the team believe it’s more functional, some believe it’s deeper-rooted emotional drivers. Can you simulate a dialogue between Carl Jung and Sigmund Freud and a Chief R&D Lead to dialogue drivers of preference and choice?”
Four
You have to use it to critique. Usually, AI is a complicit assistant, but when specifically asked and in the right way, it can be a ruthless editor.
A Quick Demo: “I’m super sceptical about this equity audit from Ehrenburg-Bass, can you read it and help me understand why my gut is telling me to approach it with caution?
Where are all the lapses in logic and ungrounded assumptions?”
Five
Rapid scenario planning by projecting into a multiverse of possible scenarios to test the validity of the strategy. Engineers call this ‘The breaking point’. You can generate a list of potential scenarios then project the strategy into those scenarios to see how it performs.
It’s really not as difficult as it looks. Once you start getting to know your AI the apprehension, misinformation and misconceptions will start to disappear. They were just a hallucination.
About the author.
Jon Williams is the Founder and CEO of The Liberty Guild