Fish Food: Episode 612
Cyborgs and centaurs in AI, Notebook LM, Stephen Fry on AI (hey that rhymes), bad dashboards, knowledge layering and TikTok creativity
This week’s provocation: Cyborgs, Centaurs and Working with GenAI
The chart above on the three stages of expertise was created some years ago by strategist and researcher Simon Wardley (who once gave a brilliant talk at a Google Firestarters event) as a bit of a joke. But there’s a lot of truth in it, particularly when it comes to complex topics like AI. As we get into the models and tools and understand more about how they work, there’s a very real hazard that we over-estimate our level of expertise (in a Dunning-Kruger kind of way).
I’ve been very conscious of this as I’ve undergone something of a journey over the past year or two with GenAI tools. I’m no AI academic but as we learn more about the topic and the tools the risk for all of us is that we think we know more than we do. I’ll admit that in the early stages of my journey there were a good few months when I struggled to really understand the use cases for how AI tools could enhance my work. The results were kind of ‘meh’. I wasn’t really sure when or how I should use them in the flow of the different things that I do.
But the stubborn part of me wanted to work it out, to understand it well enough to have an informed point of view, and to be able to draw real value from it. The AI S-curve is ramping up so quickly already, and I was determined not to be left behind. I spent time with the tools. I slowly found more and more instances of where it was helpful, what it was good at and was not good at, and how specific tools can be good at different things (I mainly use ChatGPT, a few custom GPTs, Perplexity and Gemini but Globe Explorer, for example, is brilliant when you need a structured overview of a topic - anything - you don’t know that well).
But the true unlock for me was to learn more about prompting. About the subtleties of chain-of-thought and interactive prompting and how to converse back-and-forth with the GPT to get what you need. I did some deliberate learning on this and some changes in approach, along with more thought about how I could apply it, transformed the value it has for me. As Zoe Scaman suggests in her excellent deck on Strategy in the Era of AI, seeing it as a rich conversation where you can not only summarise and search, but also dive deeper, improve outputs and categorise thinking really helped me to work out the value for me. I started to use it in a different way, using it throughout a process rather than for the odd one-off thing. The more I used it the better it got.
Earlier this week I ran the IPA’s (Institute of Practitioners in Advertising) first course in the Advanced Application if AI in Advertising. It was great. Lots of fascinating discussion about how AI is already upending the industry but also plenty of optimism around strategic application and some fun getting hands on with the tools. Physicist Richard Feynman once said that the best way to learn anything is to teach it to a child because you can only teach it well if you can simplify it effectively. And in order to simplify effectively you need to truly understand the topic. As Shane Parrish puts it: ‘Simplicity reveals a depth of understanding, while jargon often conceals ignorance’. I wasn’t teaching children of course, but the act of designing the course and thinking more deeply about application was still wonderfully useful because it helps you to structure your thinking.
Academic Ethan Mollick partnered with BCG a year ago on a well-discussed study to understand the impact of GenAI tools on productivity and work outputs. They had a bunch of BCG consultants, some of whom used the tools and a control group that didn’t. They found that consultants using AI finished 12% more tasks on average, 25% more quickly, and produced 40% higher quality results than those without. But what also came out of that study was that those using the AI tools had two different approaches, which they labelled ‘Cyborgs’ and ‘Centaurs’. Centaurs typically had a clear divide between AI and human tasks. They allocate responsibilities based on the strengths of each, handing specific tasks off to the AI to do. Cyborgs blend machine/human working, integrating and intertwining the two in the work, working more in tandem with the AI. But one approach was not necessarily better than the other in terms of outputs or impact.
And this is the point. Across what Ethan Mollick calls ‘the jagged edge’ of AI adoption and application people will take different approaches. Everyone is different but there is no one specific method. Whether you are a cyborg or a centaur finding what works for you is the only way. And that means putting the time in to find out.
The more I learn, the more I realise there is yet to learn.
Image: Simon Wardley
If you do one thing this week…
This week I played around with Google's NotebookLM which is another one of those astounding-the-first-time-you-use-it tools. NotebookLM is a note taking and research assistant - you can give it a range of different sources (from a few blog posts, to links and PDFs to whole books) and then interrogate the tool to answer specific questions about the material. You can even generate study guides and briefing docs from it. It was so good that I suspect I’ll be using it lot more in my research going forwards.
But what was particularly mind blowing was that when I gave it a few of my blog posts around the topic of navigating technological-driven change it also turned them into a super-realistic (albeit very American) 12 minute podcast format. It was incredible that within a minute I could listen to two very real sounding people discussing the main themes of my posts (you can listen to it here if you’re so inclined). It struck me how useful this can be as a learning tool given that we all learn in different ways.
Links of the week
Last week Stephen Fry gave a talk, titled ‘AI: A Means to an End or a Means to Our End?’, given as the inaugural “Living Well With Technology” lecture for King’s College London’s Digital Futures Institute. The full text is here. Lots to digest but one thing I learned about was Moravec’s Paradox which is that what we find easy the machines find hard and what we find hard, the machines easy. He quotes Donald Knuth: ‘AI has by now succeeded in doing essentially everything that requires ‘thinking’ but has failed to do most of what people and animals do without thinking …’
Good dashboard, bad dashboard is the best thing I’ve read about corporate dashboarding in a long time. (I’m going to write something on this)
I forgot to mention it last week but I wrote something on ‘knowledge layering’ - the benefits of cross-pollinating knowledge from one domain to another in the context of consulting and teaching
IBM have a bunch of new (and free) online courses in GenerativeAI - some domain (for HR, project management), others technical (NLP and language modelling)
Zoe Scaman is beginning a project on what it means to be a mother working in advertising. She’s set up a survey to get some inputs - if you’re a mum in adland you can take part here
‘Never sell your memories’. Loved this story about Cynthia Lennon and Paul McCartney
And finally…
I came across Karen Cheng’s videos via Russell Davies’ newsletter. She does some amazing films, tutorials and behind the scenes videos) on video creation using just a phone and has over a million followers on TikTok. Russell said how creativity was bursting out of TikTok and YouTube and he’s not wrong. She talks in more detail about her craft in this 30 minute conversation.
Weeknotes
This week, as well as delivering said AI in Advertising course, I was interviewing for the evolution of marketing operations models research & report I’m doing (already fascinating). I’ll be doing more of that next week. Then I’m in London and Dublin.
Thanks for subscribing to and reading Only Dead Fish. It means a lot. If you’d like more from me my blog is over here and my personal site is here. If you liked this episode do share and pass it on, and do get in touch if you’d like me to give a talk to your team or talk about working together.
My favourite quote captures what I try to do every day, and it’s from renowned Creative Director Paul Arden: ‘Do not covet your ideas. Give away all you know, and more will come back to you’.