9 companies harnessing the power of collective intelligence
By Joseph Mapue
If you’ve seen the term ‘collective intelligence’ appear in articles or online discussions, you might be tempted to assume it’s a buzzword concocted to give a new spin to the often over-hyped applications of so-called AI, Deep Learning, Big Data etc.
There are plenty of companies that use the term vaguely, or are barely beyond the idea stage, but there is also an impressive and growing group of companies that are harnessing the power of human and machine intelligence to solve the problems of tomorrow today.
Many do not rely on questionable black-box jargon, but are surprisingly pragmatic and ingenious. Here is a round-up of the ones we found most impressive.
Cognitive Edge aims to solve problems that are not adequately addressed by current primary research techniques. It is owned by the Cynefin Network, and is the brainchild of Welsh management consultant David Snowden.
Trained in philosophy, Snowden became interested in storytelling within organizations while working at IBM, who to their credit gave him various leadership roles in which he could experiment with his ideas.
His Twitter profile shows that he is a staunch anti-BS advocate when it comes to the hokum that often surrounds so-called data science in business, and makes entertaining reading!
The video above relates to Sensemaker, the flagship product of his firm Cognitive Edge, founded in 2004. Sensemaker works by ‘collecting real, unfiltered experiences on the ground to create a narrative-based repository of insights’.
The onus is placed on the person who contributes the data to interpret it, a process called self-indexing. Self-indexing ‘disintermediates’ the computer or expert, who would ordinarily act as the primary filter for the raw data.
The idea of disintermediation is that the source of data (i.e. the respondent) has the best grasp of its original context. By eliminating the interpretive layer, data capture can be continuous and feedback instantaneous.
As well as being an original idea, it is the opposite of hiding behind ‘data science’. Refreshing!
Ethelo means ‘I want’ in Ancient Greek, but also ‘I am willing’. This does a pretty good job at explaining the essence of the software, which aims to figure out not only what people want, but also what they are prepared to accept.
One of the first things you notice about Ethelo is its impressive roster of case studies, from the federal government of Canada to numerous regional organizations (mainly also in Canada) including investment firms.
The CEO John Richardson is a lawyer and a mathematician – an intriguing combination. Much of the issues that his firm deals with are related to complex, multi-faceted decisions such as how to allocate budget, or how members of a community feel about the introduction of a new initiative or bylaw.
The software is premised on the idea that traditional polling or decision-making systems suffer from a) over simplicity – issues are resolved into artificial dichotomies (Yes / No) and b) lack of engagement – e.g. low polling responses, and low voter turnout.
The Ethelo process breaks down decisions into smaller issues, elicits group responses, and uses the replies to construct a composite solution that works for all if not most.
Check out this video for an intuitive example about deciding what kind of takeaway to order.
And let’s be honest, does it get more contentious than that?
Like Ethelo, bluenove has a decent track record, claiming to have completed 800 projects in the past ten years.
In 2018 alone, it organized “70 debates for large French companies, administrative departments and local authorities”. Which using standard methods, one assumes, would have been terminally stressful for those involved. So what’s their deal?
Staffed by French and French Canadians, the company is an amalgam of three companies, and the AI component appears to have been added with the acquisition of the firm “Succeed Together” in 2019.
The Assembl software takes open-ended responses, groups them by issue, and presents them back to the participants, allowing them to comment and edit, wiki-style over the course of a few weeks.
During the process, the Assembl AI assists by identifying recurring words for real-time semantic analysis, and generating other visual aids such as mind-maps and word clouds. Users work in tandem with the AI, and can identify or classify important sentences (e.g. problem, example, actionable solution) in the combined output.
The process can go on for weeks, and the result is a set of actionable solutions that have been created, vetted and refined by the individuals they affect.
The numbers involved in individual consultations are impressive, with contributors often numbering in the thousands, and individual contributions in the hundreds of thousands. A clear case where AI assistance is required and – apparently – highly effective.
Billing itself as an “outcome-focused decision-making platform”, Loomio is a smart and pragmatic combination of existing ideas that, unusually for the field of collective intelligence, does not attempt to invoke AI in any way.
The idea behind Loomio is to create a private feed with which a closed group can interact (as if on social media) to air opinions and make decisions. The difference with traditional social media is that there are no trolls and the tools for facilitating the discussion have been expanded.
A member of a group will start a ‘Thread’ (any issue that requires discussion). Underneath, people can post comments, make suggestions and also use a variety of tools to crystallize opinion on various sub-issues (e.g. what color should the curtains be). It’s a simple idea, but well executed.
The company itself, which is based in New Zealand, lives by its values. It is a worker-owned cooperative, with roots in Occupy Wellington, and part of its original motivation was to fix the problems with bottom-up, consensus-based solutions that plagued the movement.
Powered initially by crowd-funding and subsequently by investment from social venture capital, they are increasingly being used by corporate entities (such as the Ian Martin Group in Canada).
Dr. Louis Rosenberg, CEO of Unanimous AI, has a solid resume when it comes to innovation.
His doctoral thesis formed the basis of the first use case of Augmented Reality (US Air Force), and he was also an early pioneer of Virtual Reality. While continuing his career as a technologist, he attended film school and became an award-winning screenwriter. He also writes graphic novels about transhumanism. Because why not.
His company’s main product is Swarm AI, which claims to ‘amplify human intelligence’ for both business decisions (e.g. forecasting, prioritization, trade-off analysis) and market research (e.g. product evaluations, message testing, customer feedback).
The underlying premise is that the biological principle of ‘swarm intelligence’ exhibited by bees, birds and fish when acting together in groups can be artificially recreated in human groups with the assistance of AI.
(This seems to be a recurring theme, but the company explicitly rejects the narrative that ‘AI will replace humans’, and instead places emphasis on the innate and inimitable wisdom of humans, or in this case human crowds.)
Similar to Ethelo (covered above), the software aims to create an ‘optimized outcome’ rather than a ‘most popular outcome’. The AI’s role is to observe the real-time responses of participants, assess the strength of their convictions, and suggest a new consensus to which participants can once again react. In other words, AI is fulfilling its classic role as an ‘optimizer’ rather than an innovator.
The company has also launched a sports forecasting tool (Sportspicker A.I.) that has been proven to beat Vegas in academic studies.
Dr. Rosenberg must be fun to work with.
Within the Collective Intelligence category, there is an emerging universe of e-democracy tools, which are – excitingly – being used by actual cities around the world to solve community-level problems.
Respondents submit their views in short tweet-like statements (<140 characters) and are sorted into clusters. As the process is going on, participants can view the evolving ecosystem of viewpoints and clusters graphically, and create new statements on the basis of what they see emerging.
The result of the process is a final set of ‘consensus’ and ‘divisive’ statements, which can be discussed live by the decision-makers, be they a government body or corporate board.
In Taiwan, it was used by the government to help evolve regulation on Uber (see the picture above) by involving taxi drivers, Uber drivers, Uber passengers, and other passengers in a consensus-forming discussion. It was humorously described in this talk as ‘a time when arguing on the internet went well’. That’d be a fine thing if it caught on.
In contrast to the hippy-dippy origins of Pol.is, the idea for the CrowdSmart software originally emerged out of a private equity fund that wanted to find a way to predict the likelihood of follow-on funding for individual investments.
The concept appears to be similar to an Expert Network (like GLG), only using online software to co-ordinate the process and AI to grease the wheels.
It begins with a group of contributors being asked to share their thoughts on a defined topic (an investment, a strategic move). The participants are then asked to assess each other’s responses anonymously.
According to the CrowdSmart website, the AI leverages a combination of Natural Language Processing (identify opinions), Markov Models (assess relevance, influence), and Bayesian Belief Networks (generate a predictive score). Which is a lot more explanation than most AI-based solution providers give.
The result of the process is a systematic representation of the group’s thinking. It is described as a ‘System of Record’ for each key decision.
The solution has since been used by corporations, institutional investors, and angel networks. Given the instant feedback of the ruthless investment funding process, one can only assume it works reasonably well.
According to the description on their website, the CREOpoint team seems to be a colorful bunch.
Bringing human-rights values from France, tech-savviness from Estonia and Silicon Valley confidence from the U.S. CREOpoint was founded and funded by executives from technology, media, audit, law, consulting, and psychology, including leaders coming from Facebook, EY, The Financial Times, GE, United Technologies, WPP, BNP Paribas, Dassault Systèmes and Orange.
The aim of CREOpoint in its own words is to “address the risk and spread of damaging rumors and synthetic media by crowdsourcing”, or even more succinctly as a ‘fire extinguisher for the internet’.
With the rise of cheapfake technology and the increasing prevalence of disinformation as a form of low-level domestic terrorism, CREOpoint has developed a patented software solution designed to operate quickly enough to contain outbreaks of stories that could damage lives.
The process is summarized on their website as having three stages:
1) The software identifies questionable stories relevant brands, persons and locations
2) Relevant experts are automatically polled with a question that allows them to express their confidence level in the story
3) The consensus ‘veracity score’ (and accompanying quotes from experts) is used to rapidly create a signal (article, post etc.) to either counter or boost the signal of the original story.
While the method of polling via experts may seem crude, much of the technology goes toward increasing the speed of the process without loss of accuracy. More details can be found in this detailed presentation.
The FT.com article that profiles CEO Errikos Pitsos compares Kialo to the Monty Python ‘argument clinic’, where Michael Palin goes to pay John Cleese to contradict him.
Pitsos’ actual inspiration for founding the site was his childhood, which he recalls was filled with people who liked each other having vigorous debates with no impact on their friendship. At some point, he recalls, the internet stopped being that way, and he wants to bring it back.
You can disagree with the idea, but you can’t fault the conviction of a firm that has self-funded to fifty employees.
The Kialo platform is focused purely on debating, and is designed to avoid the problems that Pitsos believes have got into the way of online discussion in the social media age. Its debates are balanced, and the points of the opposing sides concise and clear.
Within a specific topic, members create arguments for and against, which are placed separately on either side of the page, so that people can literally see both sides of the issue, rather than having to scroll through a confusing thread or email chain.
The arguments that are voted by the users as most impactful move closer to the top of the list, so that someone just entering the debate can focus on the most important points first.
The argument is protected from capture by trolls as only invited members are allowed to approve arguments. There is no AI involved, but the purpose of the system is narrow enough not to require it. It’s a logical tool that does the job it’s supposed to.
Hopefully, it makes a lot of money so that some of the other social media platforms introduce a ‘Debate’ mode.