How can we use AI to power prediction? | Source: Shutterstock

This startup is using AI to power prediction

WHEN it comes to predicting the future, we place far too much faith in the hands of experts. Despite living in a digital age where information has been made more available to the common man than ever before, the failure of accurate prediction has left many of being hit in the face with unexpected emerging events.

We just have to look at Brexit and the election of Donald Trump as examples. With both these events, the public was firmly assured by experts that neither of the results would happen.

A Singaporean startup called Scry wants to solve this problem of a growing dissonance between expectations and reality.  They have built an artificial intelligence (AI)-based platform to help businesses, governments, and institutions make smart and strategic decisions based on real consumer feedback and information.

“We want to foster greater predictability of the future, and increase our control over it through follow-on increases in decision-making efficacy,” Scry co-founder, Justin Wang, told Tech Wire Asia.

How it works

Scry is based on a philosophy that draws on research showing that non-experts with a high level of analytical and probabilistic thinking were in fact, more capable of making viable predictions than “experts”.

The Scry platform provides consumers with a platform with which to answer questions on topics that matter to them. Also known as “question engagement”, this enables a structured way to gather or crowdsource knowledge from consumers in order to make predictions.

With machine learning, users are able to gauge human predictions against the most likely outcomes in order to make better-informed decisions.

The AI-powered platform brings in research on cognition and judgment into practice on the platform.

“For example, we know from research that the higher levels of the trait “Actively Openminded Thinking” (AOT) is a strong indicator of prediction performance, so we have built in mechanics on the site that make it easy for people to be exposed to and borrow other people’s opinions,” said Wang.

“We also know that accountability and specificity increase individual prediction performance, so we have leaderboards, and scoring mechanisms that allow the individual to look back and calibrate his/her own judgment,” he added.

The platform encompasses a gamified process to engage users on the platform and allows them to track their progress. It helps users have a better understanding of major issues or trends, as well as enables them to make better decisions through a simple learning process.

“We turn forecasting into a game; you earn points and badges for actions that we know, from existing research, correlate well with accuracy. So, actions such as commenting, and replying to other users, that increase the users’ exposure to more information (on both sides of the dialogue), are recognized,” said Wang.

Discovering what constitutes a competent predictor can also be done on the Spry platform, with machine learning technology. It does this by finding out what the optimal collection of individuals or individual traits are in a given class of scenarios.

“By doing so, we can create AI models that can help our users more explicitly identify their weaknesses/areas for improvements, or boost the accuracy of a prediction by optimizing the weighting behind any group of predictions,” suggested Wang.

The AI-powered predictivity platform is not only aimed at businesses who want to make sound forecasting decisions but also individuals who want to be able to better understand what the future holds.

“People who want to learn how to get better at predicting the future can also utilize Scry’s service to help them better judge and prepare for the uncertainties of tomorrow,” said Wang.

As the algorithm behind the platform evolves, Wang believes that in the future Spry will even be able to even identify which people could be strong forecasters and good decision makers for “companies and even countries, even without them ever stepping foot into our platform before” he said.

In terms of the mechanics of Scry, Wang explained:

“Scry enables everyone to have access to its powerful crowdsourcing platform, “Our vision has always been to disseminate our predictions as widely as possible.

“We understand that for most of us, McKinsey (or similar) is not going to be on hand to provide custom in-depth analyses for the uncertainty that most organizations and individuals face.

“Unfortunately, what we have to settle with are qualitative estimates delivered through imprecise language in opinion pieces, or through the grapevine.”

In contrast, Scry enables users to sort, weigh, and estimate different parts of their argument, so instead of justifying their answer, they can fit pieces of information and opinions together under a common metric.

From this, a prediction can be made which mitigates against any unwelcome effects such as confirmation bias and scope insensitivity.






COMMENTS