- > How fast can we get answers?
- > How reliable are the answers?
- > How few people do we have to talk to?
These three points are interrelated, the common denominator being people. The smaller the sample size, the fewer respondents needed for a survey or interview, and the faster data can be analyzed and presented. But, too few respondents and the reliability, projectability and confidence suffer. Growing the sample size means higher costs and longer field times.
New survey methodologies pop up every few years to increase the efficiency of provisioning respondents. A few years ago, the buzzword du jour was “gamification”. Make surveys more fun and it won’t be so hard to get people to take them and the cost of incentives comes down. Currently, “rapid recruiting” is an active trend, enabling projects to be fielded more quickly.
At Unanimous A.I., we’re exploring a very different approach, replacing standard polls and surveys with “swarm intelligence” – a process with the potential to produce reliable insights from groups of far smaller sizes than traditionally needed. It works by changing the data collection process from assembling a set of isolated responses that are combined in a spreadsheet after the fact, to assembling a real-time group of respondents, who are connected by software into a unified dynamics system, and who work together as a “social swarm” to answer questions in synchrony.
The core of our system is a software platform called UNU. It employs algorithms modeled on swarm behaviors in nature, allowing a group of respondents to push and pull on each other in real-time as they explore a set of possible answers, converging on a solution that is not the “average response” but an “optimal response.” This reveals significant insights with much smaller groups than traditional survey methodologies.
Already we’ve seen multiple instances of the potency of swarm intelligence in which small groups of novices were able to match the predictive output of putative experts (see “How a Collaborative A.I. beat the Experts”, “Pitting UNU against the Odds Makers”).
The UNU platform also dramatically reduces the time required to find optimal answers. Dynamic-feedback loops, fueled by the real-time negotiations happening in group-decision processes, transform the contributions of individuals into a satisfactory consensus in less than 60 seconds. Consider this question that seeks to find a group consensus on the charged issue of Congress’ priorities. Traditional survey methodologies would require a battery of questions in long sequence to arrive at a group-optimized answer. Within UNU’s synchronous environment, a consensus was reached in 39 seconds.
More exploration is needed to refine how UNU fits into the market researcher’s toolbox to discern quantitative and qualitative insights. To that end, we are building a Research Partner Program. If you are a market researcher and interested in working with us to explore UNU’s potential and application to your real-world needs, send me a note.
We’re confident that UNU’s ability to rapidly focus and distill the input of individuals into actionable insights will give researchers a powerful new set of tools for examining their markets – tools that yield reliable answers from smaller samples in far shorter times. And with respondents reporting that UNU is “fun” and looking for more opportunities to participate, in the future researchers may find better places to spend their time and budgets than buying sample.
If you are interested in participating in your own swarm, sign up here.
If you’d like to read a short academic paper, see: Human Swarms