How AI Can Help Solve The Biggest Problem With Crowdsourcing

The concept of engaging “the Crowd” through digital platforms has been around for some time. Howard Rheingold coined the term “Smart Mob” in 2002 to describe the phenomenon of people acting in concert “because they carry devices that possess both communication and computing capabilities”. The concept was carried forward in 2005 by the editors of Wired to coin the term “Crowdsourcing” (crowd + outsourcing) to describe production with the a digitally connected marketplace. In the 15 years since the concept of Crowdsourcing was introduced, we have seen a wide range of crowd-based business models emerge: Wikipedia (collective knowledge), Lego Ideas (design your own kit), Kickstarter (crowd funding), Local Motors (crowdsourced vehicles), and Dell’s Ideastorm (the original social suggestion box).

With the wide range of crowdsourcing experimentation, we’ve also seen the limits of what the current platforms and practices can produce, and it isn’t pretty. Consider:

• On average, less than 30% of Crowdfunding campaigns reach their goals. On some platforms it can be closer to 10%.
• Quirky, once the darling of crowdsourced consumer goods, filed for bankruptcy in 2015.
• Dell, an early pioneer in crowdsourcing, has been able to implement only 2% of the ideas submitted on IdeaStorm.
• Independent crowdsourcing research, including a recent study by the Swiss Federal Institute of Technology, discovered that social influence can cause “herding towards a relatively arbitrary position.”

What are the key challenges?

The most common limiting factors to Crowdsourcing initiatives are one, or a combination, of the following:

1. Engaging the right crowd: Perhaps the most critical challenge in crowdsourcing is finding, and then engaging, the members of the crowd with the knowledge, skill and motivation to participate. Without domain knowledge and skill crowdsourcing produces only low quality results. Without motivation, you have unrealized potential.
2. Creating an iterative development process: One of the early corporate adopters of crowdsourcing, Dell’s Ideastorm, learned early on that creating an experience that solicits ideas without giving the community the ability to refine and evolve the ideas is a waste of time. After collecting over 10,000 ideas in the first 2 years of IdeaStorm, Dell was left with 9,750 that couldn’t be implemented, causing frustration for the company and their crowd. By introducing multi-staged challenges dubbed “Storm Sessions”, Dell was able to source and develop products with their crowd, most notably Project Sputnik, the first Linux-based laptop for developers.
3. Developing short and long-term feedback loops: The process and infrastructure required to support short-term feedback loops is difficult and labor intensive, requiring personal interactions and manual data management. Longer term feedback loops that include market data are currently next to impossible.
4. Creating intelligence from crowd data: The amount of data a typical crowdsourcing initiative produces is overwhelming, and managing this data to create knowledge and insight, even moreso. Consider the amount of manual processing and scoring overhead associated with the 25,000+ ideas in the previously mentioned Dell IdeaStorm example.

How A New Take on Collective Intelligence Can Help

Collective Intelligence, a disciplined approach to the “wisdom of the crowd”, is defined as the “science of scaling insight from multiple knowledgeable perspectives and experiences into predictions”. We’ve traditionally thought of “The Crowd” as exclusively human, but what if we expanded the collective “we” to include the rapidly evolving domain of Artificial Intelligence? The combination of expert communities and artificial intelligence is the core of a new approach to Collective Intelligence being developed by a new startup named CrowdSmart. Specifically, CrowdSmart technology creates a means to predict startup success factors by engaging an expert community of investors to score and provide critical feedback to early stage startups. Investors save time on research and improve the quality of their deal flow, and Startups get critical and timely feedback to help increase their odds for successful outcomes.

What is uniquely valuable about the CrowdSmart approach is leveraging Artificial Intelligence to detect the statistically significant ranked comments behind any given score. These ranked comments are the “drivers” that produce a specific score. The qualitative “wisdom of the crowd” becomes quantitative intelligence that grows in value over time.

According to Tom Kehler, Chief Data Scientist at CrowdSmart, “Collective Intelligence significantly outperforms individual expert intelligence at predicting the success of a new products, services and startups.” If Tom is correct, the application of Collective Intelligence will have far-reaching effects on the future of Crowdsourcing, paving the way for a more disciplined approach and more successful outcomes.

Disclosure: CrowdSmart is a Structure3C client.