Crowdsourcing, Collaborative Intelligence In Generative AI Age
The Merriam-Webster Dictionary defines ‘Crowdsourcing’ as “the practice of obtaining needed services, ideas, or content by soliciting contributions from a large group of people and especially from the online community rather than from traditional employees or suppliers.”
The definition of crowdsourcing sounds broadly like how generative AI works. Generative AI learns from data (third-party intelligence) and produces new content/solutions driven by prompts using neural network-based algorithms. In other words, crowdsourcing is exactly what generative AI excels at! So, the question is, “Will generative AI break the back of crowdsourcing?”
The other question to ask is, “Can we crowdsource the training data collection and data labelling that generative AI needs from domain experts and improve speed to market?” We could also take this thinking one step further and ask, “Can we crowdsource the skills required to rate the training data and moderate the output so that the wisdom through groups and communities is available at scale?”
The possible answers to the questions are interesting and I am sure you will want to reason out what lies ahead for yourself.
For the moment, let me seed your thinking. We know that crowdsourcing can be leveraged in several ways to speed up the progress of generative AI. In addition, crowdsourcing will solve a significant problem by bringing diversity of thinking and helping reduce bias. Once crowdsourcing around generative AI catches on, it will undoubtedly boost the gig economy.
Given that many, like the MIT Technology Review, say that “the future of Generative AI is niche, not generalised,” domain experts will be in demand. From a business perspective, one more factor weighs in for crowdsourced domain-specific skills: Generative AI is neither cheap nor easy; crowdsourcing data collection, labelling and rating of the training data can help bring down costs.
However, the first question is the most important: “Will generative AI break the back of crowdsourcing?” Is it possible that the end of this generative AI story will go down the same path as Aesop’s Fable, The Goose that Laid the Golden Eggs? Will generative AI ultimately make the wisdom of crowds redundant?
The Harvard Business School has tried to explore the “uncertainty about the role of human solvers” in a paper titled The Crowdless Future? How Generative AI Is Shaping the Future of Human Crowdsourcing. To understand the impact of generative AI on crowdsourcing, researchers at Harvard Business School launched an experiment comparing the capabilities of GPT-4 and humans. Their conclusion was not surprising.
It amplified “the promise of AI in augmenting human crowdsourcing for solving complex organisational problems.” The researchers said their conclusion “sets the groundwork for a possible integrative human-AI approach to innovative problem-solving.”
The Harvard research emphasises that little is known about Large Language Models (LLMs), which are at the heart of generative AI technology and their ability to provide novel and ingenious solutions for business problems. On the other hand, business history is peppered with human ingenuity. Without ingenuity, business would hardly advance.
More importantly, the researchers point out why using crowds to rate the training data used by generative AI will fail. They argue that increasing the number of participants in a crowdsourcing contest decreases the likelihood of winning and reduces the probability of attracting diverse participants. Besides, generative AI models cannot, at the moment at least, hope to match the abstract thinking that goes into solving business problems. The way out of this cul-de-sac lies in teaming LLMs with human strengths.
Human-machine collaboration has always been a winning combination. This is because businesses need to explore and experiment with ideas; they need to reimagine business processes and judge them for financial viability and social acceptance; finally, they need someone to take responsibility for outcomes. Only humans can do all of this. Machines provide us with a way to accelerate the process and drive efficiency.
An over-reliance on automation and AI can become a risk. It could lead to organisational complacency, increased data and privacy risks and a steady erosion of critical skills. We cannot have a world where everything changes, but no one is in charge.
The argument for embracing collaborative intelligence is strong. So, some of the work will be “crowdsourced” via generative AI engines while the higher duties of experimenting, reimagining and shaping social acceptance and accountability will be in the hands of humans. How will businesses create strategies for collaborative intelligence? How will they set KPIs for it? How will they measure the effectiveness of their collaborative intelligence approach? We need to answer these questions in the age of generative AI.
_20240501181242original_image_36.webp)