.png)
Depending on who you’re listening to generative AI is a panacea, the beginning of the end, or just the next in a long line of overhyped technologies. The truth is often somewhere in-between. Yes there’s a lot of hype, but there are also some amazing opportunities now and for the future.
Change can be scary, but it’s exciting too.
Apparently as a society we overestimate the short term impact of technological change while underestimating the long term implications (credit to Tom Goodwin for that one). I think that might be the case with Generative AI. Don’t get me wrong, Large Language Models (LLMs) like ChatGPT and Midjourney can do some incredible things and supercharge our productivity and impact.
But LLMs have significant limitations. At their heart these are probability models, predicting the most likely next word or pixel based on analysis of giant sets of training data. They’re not intelligence.
Seemingly dumb errors often occur when a statistically probably response happens to be utterly wrong. And because the high probability errors are often credible, the less knowledge you have in the space you are exploring the harder they are to spot.
Training on larger data sets doesn’t solve the problem, because it’s still using probability, not reasoning. Crafting ever more complex prompts doesn’t solve it either. Yes you’re steering the model with your own logic and reasoning, but it doesn’t stop the hallucinations.
The result of these limitations is responses prone to factual errors, inconsistencies, and a tendency toward the obvious, the average and the banal. Less like intelligence and more like the output from a room full of hundreds of inexperienced interns, churning through data to come back with information not insight (thanks for that one too, Tom). This ‘heavy lifting’ can be very useful of course, helping gather basic facts, develop hypotheses, summarise large datasets, create stimulus, draft questionnaires, guides, concepts and presentations, optimise content, generate ideas and more.
But generative AI shouldn’t just enhance existing processes and tools, it’s an opportunity to rethink the fundamentals of the way we work.
Dig deeper into the underlying technologies and we can transform how we gather, share and harness insights, adding value to the brand building process in ways hard to imagine from merely experimenting with chat models.
Yes let’s keep out feet on the ground and use these chat based interfaces to their full potential. But we need to keep our eyes on the horizon too. Setting out a broader vision for how the underlying generative AI technologies can revolutionise the insight and strategy space. That’s what we’re doing at ONE Virtual Strategy Studio.
We have nine foundational principles that underpin everything we do and push us towards achieving our goal of a 100% virtual strategy studio.
The 9 Runes of Empowerment…
1. Big ambitions – Embrace the challenge of reinventing how the biggest questions in insight and brand strategy are solved (improving existing tools and processes is not enough)
2. Industry expertise – Build how insight & strategy experts think and act into the heart of everything we create (generalist intelligence will give generalist answers)
3. Deep data – Focus on digging out the high value, high intention data that can meaningfully shape decision making (striving for ever bigger datasets won’t bring insight)
4. Repeatable frameworks – Apply analytics that go a layer deeper and can be applied and compared across data sets (factual summaries don’t answer strategic questions)
5. Scale & quantification – Run analysis at a scale that gives quant validation and an evidence trail for decision making (big businesses needs numbers, not chat)
6. High-value thinking – Ladder up to genuine insight and breakthrough thinking that solves the toughest challenges (average thinking just adds to the data pile)
7. Virtual collaboration – Get teams of intelligent models to work together, creating iterative solutions (isolated tools are hard to navigate and add less value)
8. Human relationships – Stay close to people active in the insights and strategy world and partner with them all the way (tech developed in isolation usually fails)
9. Having fun – Inject some fun and humanity into everything we do, insight & strategy can be a bit too serous (people engage with stories, not software and solutions)
By applying these principles we hope to create something truly transformative that can revolutionise the worlds of insight and brand strategy.
Not a piece of software, a virtual strategy studio you can work with like you work with agencies today. But with unparalleled speed, depth and breadth of data sources, analytical sophistication and quantified, evidence based stories, all fuelling growth for high impact brands.
