AI: Observations after 6 courses

I ran my first AI for Marketing course for the Chartered Institute of Marketing in September this year. It is the most in-demand course that I have ever run for CIM. The courses are selling out and new dates are being added to the schedule regularly. I have now run the course 6 times in 2 months, including a couple of tailored in-company deliveries. The course is designed to be an overview of what is possible, what tools you might use and what the risks are. It is not too technical and is designed for people who just need to know what all of the fuss is about and how it might impact their marketing plans going forward. You can find out more on the CIM web site.

The attendees have ranged from Marketing Directors to marketing graduates from almost every industry sector you can think of. Some have worked for agencies and tech companies, but most have been client-side marketers who are responsible for managing campaigns.

The course covers AI and its impact across the marketing spectrum, including:

• Generative AI for creating content (copy, video, images etc)
• AI for campaign management (search, social media, email etc)
• AI in customer service (chat bots, personalised recommendations etc)
• AI in data analytics (predictive modelling, content testing etc)
• AI and its impact on newer digital channels (AR, VR, metaverse)
• The ethics and risks of AI

Since I started to research and deliver the course, I have learned a lot. here are my initial observations:

AI is not perfect and you need human intervention

• It will hallucinate. It will make stuff up. I asked ChatGPT "Who is Nick Baggott?" most of its response was accurate, but it described me as a successful author of many digital marketing books. Alas I have never written any sort of book. Check everything - the facts, the style, the tone....

Generative AI only starts to work well when you brief it well

• You have to think of using AI as if you are briefing an agency. The better your brief, the better its response. Your Prompt Engineering is key to this. Tell it what style you want it to write in, what to include, what not to include and how long it should be. For example, ask it to include 3 academic references or write in a professional thought leadership style. You can have fun and ask it to write in the style of Donald Trump or Walt Disney if you like.

Data is a key enabler and the usual rules apply

• AI tools such as Dataiku can build predictive models that will help you to recommend the right products or display the right content. You can turn a simple web site A/B content test into a complex on-going experiment where machine learning drives countless simultaneous tests. But as always, it is garbage in and garbage out. You need to put in clean data and have a single view of your customer before this can really start to perform.

AI is not just ChatGPT

• There are some wonderful visual content tools like Dall-E that will generate original images (like the cover picture of a koala riding a bike).
• There are video creation tools like Synthesia that will turn your slides into a video (probably best for internal rather than external use as it still looks a little robotic).
• There are great tools to help with ChatBots, Search Marketing, Social Media Listening and so much more. You'll have to come on the course to find out about all of these and more!

AI is here to stay

• We have some great ethical discussions concerning inclusivity, machines taking the jobs of humans (where have we heard that before?), the risks of fake content and hallucinations, data privacy and so much more. But, on balance, pretty much everyone has said that if it is used properly, the benefits of AI outweigh the risks.

I would love to hear your observations and experiences of AI so far. Please let me know in the chat what you have learned so far and what you would love to know. if you want to attend a course, you can follow this link to the CIM or send me a DM and I can arrange a 1:1 or discuss a bespoke course for you and your colleagues.