The Benefits of Crowdsourcing Innovation with Artificial Intelligence and Machine Learning

In exploring the benefits of using AI to generate ideas, Ricardo Garza of Softtek explains that one of the most intriguing aspects of neural network technology is its ability to discern patterns that humans overlook.

Businesses have long recognized that the best and most innovative ideas usually come from the front lines – originating from the individuals directly involved in manufacturing products, executing processes, and overseeing operations.

For large organizations, however, the task of soliciting, reviewing, and considering massive volumes of ideas quickly becomes overwhelming. Specifically, how do you separate the truly creative and inspired proposals from the crowd of also-rans? And, once identified, how do you ensure that the best ideas get implemented quickly?

Online idea generation platforms that allow users and managers to evaluate innovation proposals can help. These platforms can incorporate interactive gaming applications to invite participation and circulate feedback. Real-time translation systems can help large, multi-lingual enterprises easily share ideas.

Still, identifying the best ideas and then assessing their viability in a timely manner presents a management challenge.

Idea Acceleration

Today, artificial intelligence and machine learning capabilities are being applied to develop neural networks that dramatically streamline the task of innovation management. Sentiment analysis capabilities gauge how peers respond to a suggestion. For example, does the idea spread rapidly? Is feedback enthusiastic? Contextual analysis and pattern recognition programs, meanwhile, can suggest related topics or approaches in response to an idea submission.

Ensure that great ideas — rather than becoming lost in the crowd — pop to the forefront.

A large organization can use the smart networks to solicit and evaluate ideas and proposals for innovations ranging from process improvements to customer promotions to store displays. Algorithms apply advanced clustering techniques to aggregate data from submitted proposals and score each proposal’s probability of success. Criteria that the algorithms consider include how many people look at the proposal, how many people comment on it, and how quickly and broadly the idea is shared across peer groups. Pattern-recognition capabilities identify proposals offering similar or related ideas.

Through this process of data collection and analysis of multiple criteria, machine learning-based neural networks can drive “idea acceleration” and speed the process by which crowdsourced innovation proposals are evaluated and implemented. This can help ensure that great ideas — rather than becoming lost in the crowd — pop to the forefront.

A Better Deli Experience

To look at a real-world example, there was one North American grocery chain that created a new omnichannel strategy for the chain’s retail deli operations. The objective of the initiative was to enhance the customer experience, specifically by reducing the time that customers spent waiting in line to place and receive their orders. Ideas for innovating the in-store experience were crowdsourced from customer-facing staff as well as development teams.

Machine learning and cognitive capabilities were deployed both to select and capture submitted proposals, as well as to evaluate and promote ideas that were most likely to succeed.

One of the key distinctions between rule-based systems and machine learning (ML) is predictability of outcomes.

The project team originally planned to develop a downloadable mobile app for customers to use to place orders. But the proposals analyzed by the cognitive platform suggested a different approach: Improve the customer experience through direct engagement.

Rather than requiring customers to download an app — a barrier to entry — the team developed a mobile app for store personnel to use to facilitate the ordering process. As customers enter the deli area, staff members approach the customer to take orders and capture the customer’s mobile number. When the order is finished and ready for pick-up, the customer receives a text message (with the order number) directing them to the pick-up counter.

And because the app is integrated with the deli department’s point of sale (POS) system and its inventory database, this enables ongoing data analytics to enhance supply management and sales insights.

The results speak for themselves. Within six weeks of deployment, the redesigned process accounted for 15% of the deli department’s sales. The original target was 2%.

Rules vs. Neural Networks

One of the key distinctions between rule-based systems and machine learning (ML) is predictability of outcomes. A basic robotic process automation (RPA) tool executes a specific task according to specific rules. Outcomes derive precisely from the rules the bot has been taught to follow and can be traced back to those rules. In other words, there are no surprises.

One of the most intriguing aspects of neural network technology is its ability to discern associations and patterns that humans overlook.

A machine learning-based neural network, meanwhile, uses algorithms and features engineering criteria to process structured and unstructured data, apply pattern recognition capabilities, and identify causal relationships.

The users supplying the data and applying the algorithms often can’t explain how or why a neural network comes up with a particular conclusion. Indeed, one of the most intriguing aspects of neural network technology is its ability to discern associations and patterns that humans overlook.

Today, a growing field of neural network research is focused specifically on understanding how neural networks make decisions. As we continue to evolve and this landscape develops further, insight into this question will likely drive significant advances that will help companies improve their operations and implement innovative ideas.

(Image credit: geralt / Pixabay)