Hunter Thurman
Founder and President, Alpha-Diver
www.alpha-diver.com
hunter@alpha-diver.com
How should we be thinking about experimenting with AI?
AI is dominating the conversation among insights teams these days. But before jumping into experimentation, we need to reframe how we think about AI. This emerging technology won't inherently unlock new understanding – the inputs we feed it will.
Rather than "artificial intelligence," think "all-inputs." AI is only as smart as the data we provide.
Remember the fizzle of big data? Behind the hype lay more flawed, mediocre inputs. More data ≠ better data. And this provides a cautionary tale beyond the hype of the AI hysteria.
Would you prompt ChatGPT to "write a market research report" and publish the generic output? Of course not. The quality reflects the inputs.
As Professor Richard Baldwin famously put it, "AI won't take your job, someone using AI will." AI is a tool to enhance our work, not replace it. Like digital in the 2000s, AI can help us work smarter – if we use it right. In a recent webinar, Insights Association CEO Melanie Courtright observed that the transition to AI closely resembles the transition to digital in the early 2000s. The words, promise, ideas and hysteria feel reminiscent of that period.
But, just as with digital, AI starts with the same old challenge: clearly defining the problem. And aligning stakeholders. And communicating simply. AI augments human intelligence by tackling tedious tasks, not by solving problems for us.
The victors will be those who focus first, not on the technology, but the inputs. Unique, insightful inputs – not the same vanilla data repackaged.
So before launching new AI experiments, scrutinize your inputs. Are they capturing true human behaviors and motivations? Or just weak proxies and assumptions? The quality of outputs depends on it.
AI's full potential emerges when combined with richer inputs from new data sources and advanced methodologies. Don't just dust off and digitize a bad survey. Use AI to mine the gems hidden in both structured and unstructured data.
The future belongs to those who feed AI something new. The same inputs will get the same outputs, no matter how slick the tech. Smarter AI starts with braver inputs.