Collective Intelligence + Artificial Intelligence + Strong Processes&Behaviors = Super Intelligence
In 2001, I read about electronic brainstorming technology, a computerized meeting system developed by the University of Arizona and IBM. I arranged for a demonstration at the 40-seat facility in the basement of the university’s business school, where I learned how electronic brainstorming helps to increase a group’s collective intelligence by countering many of the problems with traditional ways of meeting. For example:
- By enabling everyone to simultaneously type their responses to a question or request, it prevents one or two people from dominating the discussion.
- By enabling anonymous entry, it encourages contributions from those who fear that their ideas will be criticized.
- By using sophisticated voting tools, it enables the group to use its collective judgment to select or rank order ideas and alternatives.
Not long after my visit, I bought an electronic brainstorming system and launched a meeting facilitation service. I was excited by the possibility of using this technological intervention to boost a group’s IQ. Especially exciting was the prospect of enabling a group to think of ideas that none of the group members previously possessed or could have thought of alone. (That’s a sentence worth reading twice.)
Now, with the advent of generative AI tools like ChatGPT, I’m even more excited about using technology to increase intelligence—this time by combining the collective intelligence of a computer-supported group with the artificial intelligence of generative AI to create a cyborg-like entity that exhibits super intelligence—an intelligence that generates ideas that a group of humans could not have generated alone.
But there’s more to making a group smarter than supporting them with technology. The group also needs some method to its madness. I provide this in electronic brainstorming meetings by using the technology to implement three types of processes—meeting processes (e.g., a well-crafted agenda), group processes (e.g., the nominal group technique), and task processes (e.g., a strategic planning process). In this way, I combine a technological intervention with process interventions.
But even this combination isn’t enough to boost a group’s IQ because there remains the problem of dysfunctional group behaviors like wheel spinning, repeatedly wandering off on tangents, and unhealthy forms of argument. That’s where the methods facilitators use to control group behaviors (behavioral interventions) come into play. Another, and key, form of behavioral intervention is assembling a cognitively diverse group in which each of the members possesses a piece of the puzzle.
To summarize, a group is made collectively more intelligent by combining technological interventions, process interventions, and behavioral interventions. So far, I’ve described these interventions in the context of electronic brainstorming technology. The same kinds of intervention apply when artificial intelligence is added to the mix.
In this case, generative AI is the technological intervention. The most fundamental process intervention when using generative AI is prompt engineering. Prompt engineering refers to the process of crafting the input prompts—i.e., the questions, commands, and context—that are used to communicate with AI systems, especially those based on natural language processing (NLP) like ChatGPT. The way a prompt is phrased can significantly impact the quality of the response. The process of iteratively refining a series of prompts can improve the quality even more. Iterative refinement is like steering the AI system toward answers that are progressively more accurate, relevant, insightful, and useful. But steering AI without a map gets you only so far. The real potential comes from using creativity, problem-solving, strategic planning, and other processes to plot the search for answers.
Now imagine a group engaged in collaborative prompt engineering—say, a twenty-person, cognitively diverse group using electronic brainstorming technology and a problem-solving process to jointly craft and iteratively refine a series of prompts. As often happens in a group like this, the group members stimulate each other to consider ideas (prompts) in directions they wouldn’t have considered on their own. Here, the AI also stimulates the group members to explore new paths of thought. In combination, the collectively intelligent group, the artificial intelligence, and the processes constitute a super intelligent cyborg capable of generating ideas that the group of humans would be unable to generate alone.
In his book Reinventing Discovery: The New Era of Networked Science, Michael Nielsen, a pioneer of quantum computing, describes a chess tournament that allowed humans and chess-playing computers to compete together as hybrids, i.e., as cyborgs. The entrants included:
- A Hydra computer alone. At the time, the Hydra series of chess computers had never lost a game to any human chess player, including easy defeats of several Grand Masters.
- Several teams consisting of a Grand Master and lesser-ranked colleagues using a strong chess-playing computer.
- A team consisting of two low-ranked amateur chess players using three off-the-shelf computers running standard chess-playing software programs.
How do you think the entrants fared? Perhaps unsurprisingly, the human-computer teams trounced the lone Hydra computer. But to everyone’s surprise, the two amateur players using off-the-shelf computers and software beat the teams consisting of a Grand Master, lesser-ranked colleagues, and a strong chess-playing computer. Why? Because the amateurs used a better process and engaged in better behaviors—they practiced beforehand, studied their opponents, continually communicated with and supported each other during the tournament, and ultimately relied on their intuition to decide each of the moves. As the former World Chess Champion Garry Kasparov put it:
Weak Human + Weak Machine + Better Process beats
Strong Human + Strong Machine + Inferior Process beats
Or as I’m inclined to put it:
Collective Intelligence + Artificial Intelligence + Strong Processes & Behaviors = Super Intelligence