AI-Augmented Collective Intelligence 1/2025

HNY! The fourth quarter of 2024 (and a bit of January) was full of progress and insights. Below is an overview of my writing, with a link to the full articles.

1. Your Problem-Solving Idea Flow, AI-Augmented

2. Harden Your Ideas with AI

3. Superhuman Knowledge Workers? AI Exoskeletons and Scaffoldings

4. What’s GenAI Really Useful For - Update

5. Four Drivers of Work Evolution

6. From "Me Automated" to "Us, Augmented"

Enjoy!

Your Problem-Solving Idea Flow, AI-Augmented

TLDR: How do we use AI to augment our capabilities to generate better ideas and solve problems? This deep dive builds on the previous posts and gives practical "how-to" guidance to anyone who wants to harness GenAI's power and up their (human) game.

🗺️ First Step: Exploring the Why

• Falling in Love with the Problem: Exploring the Problem Space (Emphasizes falling in love with the problem before finding solutions; AI can help refine understanding and avoid miscommunication.)

• Steps to Discover the Why: Confirm Problem's Description: Use AI to refine and clarify problem definitions. Personas' Perspectives: Map stakeholders and take different perspectives. Decompose the Problem: Break down the problem into parts and abstractions. Inject Doubt: Use AI to critique thought processes and identify gaps. Summarize and Filter: Cluster ideas and filter out non-promising perspectives.

🔎 Second Step: Exploring the What - Guiding AI to Interesting Solution Spaces

• Discovering the What: Understanding the category of problems and finding creative lenses.

• Steps to Explore the What - Analogize: Use analogies to generate insights about the problem. Use Lenses: Apply various organizational and cognitive lenses to gain new perspectives. Inject Doubt: Challenge the analysis to identify gaps. Work with Constraints: Force the thought process to navigate constraints. Summarize and Filter: Use AI to organize ideas and filter out unimportant thoughts.

⚙️ Third Step: Exploring the How - Co-creating with Machines

• Solution-Finding with AI as a Thought Partner: Ensure active human involvement and avoid treating AI as an oracle.

• Steps to Explore the How - Generate Solutions: Iterate ideas using insights from previous steps. Use Management Frameworks: Apply frameworks to generate solutions. Augmented Collective Intelligence Pillars: Identify ideas related to collective intelligence. Tech Use: Leverage specific technologies tied to organizational design. Recombine Ideas: Combine ideas to form alternative solutions. Work with Constraints: Constrain solutions to fit resources or requirements. Filter Ideas: Emphasize creativity, practicality, or novelty.

💪🏾 🦾 Enlisting the Full Power of Humans and AI to Push Hard

• Human in the Loop: Keep humans actively engaged, giving feedback to avoid drift. Using Different Agents: Build AI agents with specific capabilities for different tasks. Asynchronous Batch Processing: Branch out ideas and provide results at intervals. Recursive Loops: Problem-solving and creativity require revisiting multiple steps.

Harden Your Ideas with AI

TLDR: Can you catch your ideas' (and solutions, concepts) blind spots before it is too late? GenAI can now give that job a hand. Big and small ideas are rarely perfect when they first come out.

The best idea creators and problem-solvers have solid processes for fast and cost-effective idea-improvement processes based on ruthless and insightful feedback at scale. 

What if you could enlist Generative AI to help you do the same? This article explores how to do that. One insight here is that machines can critique our ideas more effectively if guided to use the methods humans have used for hundreds of years to poke holes into all sorts of things.

I also present an already useful, freely accessible functioning prototype inspired by recent work, including that of our MIT CCI Design Lab, which can be found on OpenAI’s GPT Store.

Superhuman Knowledge Workers? AI Exoskeletons and Scaffoldings

TLDR: The GenerativeAI commentary is too focused on technology. What I see about "human augmentation" is often too generic or high-level. But the opportunity is here and now: to control, harness, and compete with AI (and AI-powered organizations), we can make knowledge workers more effective and efficient, and more satisfied - individually and in teams.

💡 AI can help us do that, and that help comes in two main types.

1. “scaffolding” AI, which supports skill-building and gradually reduces assistance,

2. and “exoskeletons,” which provide continuous enhancement.

Leaders and their HR partners now have a significant opportunity to develop and deploy quick, personalized scaffoldings and exoskeletons for many roles in their organizations. Domain experts with low technology skills can now help build some. This can impact productivity and engagement and reduce toil.

However, all this will not amount to much if we can't give employees more space and reduce clutter—productivity and energy at work depend on that.

The article describes the basics of developing and executing a human augmentation strategy through scaffoldings and exoskeletons.

What’s GenAI Really Useful For - Update

TLDR: Many are still confused about Generative AI's use cases. That's also because we often see long lists that are hard to remember. And this matters: much of GenAI’s success depends on bottom-up adoption, where people close to the business domain “make it happen.”

Building on many others’ observations, I have landed on this simple guideline:"GenAI is often useful when a horde of varied, increasingly competent, and mature - but junior - employees would add value to you and your teams." That means that generative AI is an obvious choice for tasks where you...

💡 ...need access to many ideas and perspectives, across multiple knowledge areas;

📈 ...are happy with some variability in the precision of the output, or you can verify the output easily;

🏎️ ...want speed and scale;

🔎 ...can kick tires on the transparency of the logic yourself (or someone/something else);

📀 ...believe the machine has enough data (or you can feed it);

😓 ...want to spend more effort than what you would be able to alone;

🎧 benefit from ubiquitous availability across modes (text, video, sound).

You can build a scorecard for your use cases with this. This article gives you more examples and details on how to apply these principles.

Four Drivers of Work Evolution

I believe there are four main vectors of how AI will change how we work, our processes, and our own usefulness as professionals. We can plan around them.

👩🏽💻 Today, information technology primarily augments work through varying degrees of automation. This includes workflows, generative AI, and predominantly predictive AI from the previous generation, which is still being embedded and developed. Many are still figuring out how to use these machines. The future:

🤖 Driver 1: Machines Will Do More: Starting from the bottom, machines will take over tasks that, within current processes, are increasingly within their capabilities. These tasks will continue to expand as machines grow more capable.

✏️ Driver 2: Process Changes Enable Machines Do More: This means progress will not rely solely on machines becoming smarter but also on designing processes that better suit the capabilities of today’s machines. For instance, new human-in-the-loop or machine-in-the-loop systems for quality control.

🚀 Driver 3: We Will Do More of What We Do Today: We will see a significant increase in what we already do today. Current processes and services will become more affordable, enabling greater scale. For example, we’ll produce much more software code and content—text, multimedia, and beyond—which will improve overall output and productivity.

💥 Driver 4. We Will Do Things We Don't Do Yet...provided we have the necessary resources, such as energy. Looking back 100 years, many things we now do and desire were unimaginable. Similarly, we should expect new, unforeseen pursuits to emerge, and scaled supply follows.

As a result, I anticipate a profound shift in organizations and people. Beyond the increased presence of machines working alongside humans, the sheer volume of work will also require more - and new - human involvement. For instance, while machines will write more code, we will still need developers to identify the right problems to solve and quality control.

🕸️ We will likely see much more dynamic, new networked processes and organizational structures. These structures will allow agents to connect in ways that redesign processes more fluidly than current management practices permit. Some of these organizational forms may resemble markets, where demand and supply are balanced at a granular level. In such scenarios, agents—whether human or machine—could leverage resources to achieve increasingly diverse objectives.

Despite these possibilities, the exact role of humans remains unclear. My sense is that people will focus more on the "why" and the "what," becoming problem-seekers and helping machines scope problems effectively rather than serving as the executors of the "how." Many of us are unprepared for this shift; our current training infrastructures cannot support it - and our organizations don't learn enough. I have written about it extensively elsewhere.

The time to start preparing is now.

From "Me Automated" to "Us, Augmented"

TLDR: What did we learn this year about where and how to use AI to augment work? So much has happened in the last year or so that it is easy to lose track of the big insights. So here's a white paper—titled "Us, Augmented"—that collects much of my recent work. Some a-ha's that make a real difference:

❶ Using AI to help humans become good beginner learners at everything

❷ Adopting a different approach to enhance the capabilities of experts, as opposed to those of beginners – thus applying AI differently to both ends of the capability distribution curve

❸ Human “3P” skills (personal engagement, process, and prompting) when engaging productively with AI

❹ Humans’ role in the transformation, both individually and in groups, shifting from “doer of the how” to “leader of the why and what” and manager (critic and quality controller) of the machine output

❺The deliberate thinking process of “human(s)+AI(s),” discovering the problem space, exploring possible solution’s categories, and, later, iterating and refining solutions

❻ Using constraints and concepts derived from existing, human-made frameworks

❼ The value of using AI machines to help humans recombine ideas

❽  Identifying how to use AI not just by comparing capability with humans but also in terms of differential ability to expend effort

❾ The current unnecessarily comparatively undeveloped landscape of quality management methods and AI leadership (as compared to people leadership) practices

❶⓪ The critical importance of using AI well to improve people’s ability to function collectively in their networks

❶① The need to guide workforces proactively in developing the right skills, but also their future career narrative to avoid systemic resistance to change

❶② The imperative to reimagine our learning infrastructures away from point training and toward a continuous flow of peer-supported in-the-flow-of-work skill development

❶③ The opportunity afforded by using groups of AIs combined with groups of people and the need to focus on that vision to design AI-powered tools and processes

❶④ AI is a “system 1” thinking complementing humans’ “system 2” through exoskeletons and scaffolds, harnessing the recombinations from very different fields of quality control

❶⑤ More broadly, the opportunity to use AI well inhuman networks to make “the world know what the world knows” and boost our knowledge-based economies and societies.

These and other insights find practical application in designing organizational processes and structures, as well as evolving skills, knowledge management, and collaboration infrastructures. These are “high-leverage points” in the system dynamics of companies, organizations, and ecosystems.

We can design and build them. Enjoy the read, and share liberally. All feedback is invited - this is how we make this field grow.

This newsletter is part of a series on AI-augmented Collective Intelligence and the organizational, process, and skill infrastructure design that delivers the best performance for today's organizations. More here.

Get in touch if you want these capabilities to augment your organization. Build learning, problem-solving, innovative, intelligent organizations. Build Superminds.