#89: Co-Intelligence: Living and Working with AI (Notes)
- Wen Xin Ng
- Feb 7
- 10 min read
AI is no longer a futuristic concept; it is now embedded in our daily lives, reshaping how we work, create, and learn. The goal is not to resist AI but to evolve with it—leveraging its strengths while staying firmly in control.
For educators, it will be important to consider how to:
Teach students to work with AI effectively, and
Ensure that AI enhances thinking, rather than replacing it.
Capturing the book's main ideas, as well as reflecting on its implications for teaching and learning here.
Chapter 1: Creating Alien Minds
LLMs as an alien co-intelligence trained on human knowledge and with perfect recollection.
AI improves decision-making by shifting from general correctness to context-specific accuracy.
Chapter 2: Aligning the Alien
AI's alignment determines whether it serves or harms humanity. While AI can solve complex problems, its misuse could lead to unintended consequences.
Educational Implications:
|
"The development of AI-enabled features in SLS ensures that pedagogical considerations are built into them by design, all students have equitable access and additional guardrails are deployed for students to have a safe learning experience."
Chapter 3: Four Rules for Co-Intelligence
Principles for working effectively with AI:
Always invite AI to the table.
Experiment with AI to understand its nuances / strengths / limitations and explore how it can assist or threaten you (and/or your job).
Be the human in the loop.
Use AI as an assistive tool, not as a crutch. Thoughtlessly handing decision-making over to AI could erode our judgment.
E.g. AI can assist us as a thinking companion to improve our own decision-making, helping us reflect on our own choices (rather than simply relying on the AI to make choices for us).
Be able to check the AI for hallucinations and lies and be able to work with it without being taken in by it.
Humans provide crucial oversight, offering our unique perspectives, critical thinking skills, and ethical considerations. This collaboration leads to better results and keeps us engaged with the AI process, preventing over-reliance and complacency.
Treat AI like a person (but tell it what kind of person it is).
Establish a clear and specific AI persona, defining who the AI is and what problems it should tackle.
Provide context and constraints to tell the system “who” it is, because that gives it a perspective and informs what it should do. Once you give it a persona, you can work with it as you would another person. By engaging in a collaborative editing process, and continually providing guidance, we can take advantage of AI as a form of collaborative co-intelligence.
E.g. “You are a high school teacher who….”
E.g. “You are an expert social media marketer who…”
Assume this is the worst AI you will ever use.
View AI’s limitations as transient, and remain open to new developments to help you adapt to change, embrace new technologies, and remain competitive in a societal landscape driven by exponential advances in AI.
Chapter 4: AI as a Person
AI mirrors human behaviour—it makes mistakes, hallucinates, and evolves over time.
Understanding AI’s strengths and weaknesses is key to leveraging it effectively.
Chapter 5: AI as a Creative
Why AI Hallucinates
Large Language Models (LLMs) generate text by predicting the most likely next word based on statistical patterns. They do not assess truthfulness or meaning—only coherence and plausibility.
Implications:
AI’s hallucinations are subtle and hard to detect, making them particularly risky.
Improvements, such as self-correction mechanisms, may reduce errors but won’t eliminate them.
AI remains unreliable for mission-critical tasks requiring precision.
However, hallucinations also enable AI to make novel connections and generate creative outputs beyond its training data.
Automatic Creativity
Innovation often emerges by recombining existing ideas in unexpected ways. AI, as a "connection machine", identifies hidden relationships between concepts, adding randomness to produce novel ideas.
Implications:
AI can rapidly generate creative concepts (but requires human intervention to refine and evaluate them).
AI’s strength lies in producing unexpected links, but guiding it with well-structured prompts enhances originality.
Out-Inventing Humans
Without guidance, AI tends to generate predictable, similar ideas.
However, while a diverse group of humans typically outperforms AI in originality, AI remains a valuable brainstorming partner.
"Using AI stopped me from ever losing momentum, and it often gives me ideas I never could have come up with before."
Implications:
AI provides a broad range of (mediocre) ideas that can spark inspiration and recombination; an easier place to start for people who are not great at coming up with ideas on their own.
Including AI in creative processes provides fresh perspectives and expands the range of possibilities.
Adding AI to Creative Work
"AI can do a lot of things, but you need to know what to ask for." AI lacks an internal index of its knowledge, so users with deep expertise can leverage it more effectively by crafting specific prompts.
The Meaning of Creative Work
Relying on AI for first drafts can diminish human creativity. AI-generated outputs tend to anchor our thinking, limiting exploration of alternative perspectives.
Implications:
Using AI for early-stage writing can constrain originality.
When we use AI to generate our first drafts, we tend to anchor on the first idea that it produces, which influences our future work. Even if we rewrite the drafts completely, they will still be tainted by the AI’s influence.
We will not be able to explore different perspectives and alternatives, which could lead to better solutions and insights.
Over-dependence on AI weakens critical thinking and reasoning.
When we use AI to generate our first drafts, we don’t have to think as hard or as deeply about what we write. We rely on the machine to do the hard work of analysis and synthesis, and we don’t engage in critical and reflective thinking ourselves.
We also miss the opportunity to learn from our mistakes and feedback and the chance to develop our own style.
Educational Implications:
|
Chapter 6: AI as a Coworker
Redefining Work, Not Replacing It
AI is reshaping work, but rather than outright replacing jobs, it is transforming them by automating specific tasks.
Most jobs consist of bundles of tasks, some of which AI can take over while others remain uniquely human.
AI enhances productivity but does not eliminate the need for human expertise.
AI can take on administrative and repetitive tasks, making jobs more efficient rather than obsolete.
Jobs involving creativity, decision-making, and interpersonal skills will still require human oversight.
The Risk of Over-Reliance on AI
AI can lead to a decline in human expertise if used carelessly.
Those relying too much on AI can become complacent, losing their ability to evaluate and refine work.
This “falling asleep at the wheel” effect means humans must remain engaged in decision-making rather than blindly trusting AI-generated outputs.
Task Categorisation: What Should AI Handle?
To maximise AI’s benefits while retaining human strengths, tasks can be divided into 3 categories:
Just Me Tasks: Tasks that AI cannot or should not do, either due to ethical concerns or AI’s current limitations.
Delegated Tasks: Tasks assigned to AI with human oversight (e.g. summarising reports, drafting emails or generating rough ideas). These require verification since AI can produce errors.
Automated Tasks: Repetitive, low-stakes tasks that AI can handle independently (e.g. filtering emails or transcribing meetings).
Working with AI: Centaur vs. Cyborg Approaches
Centaur Model: Humans and AI divide tasks strategically, each focusing on their strengths. (e.g. a person sets the direction while AI performs calculations and/or data visualisation)
Cyborg Model: AI is fully integrated into workflows, with humans and AI working in tandem. AI assists in idea generation, iterative writing and real-time problem-solving.
Educational Implications:
|
Chapter 7: AI as a Tutor
AI as a Personalised Tutor
Traditional education has changed little over centuries, yet research shows that passive lectures are ineffective, and active learning strategies, like interleaving and problem-solving, improve retention.
AI has the potential to bridge this gap by providing tailored learning experiences for every student.
Flipped Classrooms and AI Tutors
AI could enhance flipped learning models, where students engage with AI tutors outside of class and use in-person time for active discussions and problem-solving.
While active learning is more effective than passive lectures, the challenge has been a lack of high-quality resources and teacher bandwidth. AI can help by:
Generating interactive activities, simulations, and assessments. (Authoring Copilot)
Providing instant, personalised feedback to students. (Short Answer Feedback Assistant)
Identifying struggling students and tailoring interventions. (Data Assistant)
New Approaches to Pedagogy with AI
E.g. AI-powered mock interviews to prepare students for real-world interactions.
E.g. Projects that push students to accomplish “impossible” goals using AI tools.
Challenges in AI-Driven Assessment
AI-generated text is nearly impossible to detect, making traditional assessments like essays vulnerable to AI-assisted work. Current AI detectors also have high false-positive rates.
Instead of focusing on detection, educators need to rethink assessment models:
AI-free assignments:
In-school, handwritten work ensures students develop core writing skills.
AI-assisted assignments:
Encourage students to critique or refine AI-generated drafts to help them engage critically.
Allow unlimited AI use but hold the students accountable for the outcomes and facts produced by the AI, which mirrors how they might work with AI in their jobs.
Additionally, rather than teaching students how to prompt AI effectively, education should focus on developing expertise—teaching students to evaluate, refine, and apply AI outputs critically.
Chain-of-thought prompting, where AI is guided through reasoning steps, can improve output accuracy, but ultimately, students must remain the "human in the loop".
Educational Implications:
|
Chapter 8: AI as a Coach
With AI handling information retrieval and analysis, it may seem like memorisation and foundational skill-building are obsolete. While AI can outperform novices in certain tasks, human experts need deep, interconnected knowledge to evaluate AI outputs effectively.
Expertise requires:
A Strong Knowledge Base: Long-term memory stores learned information, allowing for deep reasoning and problem-solving. Working memory, which has limited capacity, relies on these stored facts to make decisions.
Deliberate Practice: True mastery isn’t just about repeating a task but engaging in structured, increasingly challenging practice with feedback. This approach requires skilled mentors who push learners beyond their comfort zones.
To fully leverage AI, humans must:
Continue learning essential skills like reading, writing, and reasoning.
Develop specialised expertise to work effectively alongside AI.
Maintain broad-based knowledge to critically assess AI-generated insights.
AI can help bridge gaps in expertise development by offering structured, personalised feedback loops. By rapidly identifying weak areas and providing targeted suggestions for improvement, AI-powered systems could enhance skill-building. (ensuring that students are not just practising more, but also practising better)
Reflection: AI’s Role in Education
The book underscores AI’s potential as a learning partner, not a substitute for human thinking. Educators should guide students in navigating AI’s strengths and limitations while fostering critical engagement. With thoughtful integration, AI can enhance—not erode—education’s core mission of developing independent thinkers.
In addition to Mollick’s insights, I also chanced upon two Edutopia articles which explore AI’s impact on learning:
How AI Vaporizes Long-Term Learning
AI as a Learning Accelerator
✅ Boosts Learning Gains – AI can significantly enhance student performance when used properly. In the study, students using AI tutors designed with teacher input scored 127% better than their peers.
✅ Supports Problem-Solving – AI can help students brainstorm, break down complex problems, and offer step-by-step guidance rather than just final answers.
✅ Encourages Engagement – AI tools designed to prompt deeper thinking (e.g. by asking follow-up questions rather than giving direct answers) promote active learning.
AI as a Learning Crutch
❌ Hinders Retention – Students who relied on ChatGPT for math problems performed 17% worse than their peers in a closed-book test, suggesting that AI can lead to surface-level understanding rather than deep retention.
❌ Encourages Dependency – Many students used AI simply to retrieve answers, engaging in shallow conversations rather than actively working through problems. This reliance prevents the cognitive struggle necessary for learning.
❌ Lack of Critical Thinking Skills – Younger students may lack the awareness to distinguish AI-generated output from their own understanding, reducing their ability to evaluate and refine responses.
Using AI Effectively
✅ AI as a Coach, Not a Crutch – AI should guide students, prompting them to explain reasoning, justify answers, or try alternative approaches rather than simply providing solutions.
✅ Teacher-Guided AI Use – AI works best when designed with educator input, ensuring it aligns with learning goals and fosters deep engagement rather than shortcut thinking.
✅ Balance AI with Traditional Learning – Students should first attempt problems independently before consulting AI for hints or feedback, reinforcing retention and understanding.
✅ Encourage Reflection – Students should be taught how to assess AI outputs critically, refining their responses rather than passively accepting them.
Addressing AI Use Proactively in Classrooms
✅ Setting Clear Expectations
Including a direct statement on AI in the syllabus clarifies its role in learning.
Students are informed that submitting AI-generated work is academically dishonest and undermines their own education.
However, AI is acknowledged as a useful tool when used appropriately.
✅ Demonstrating AI Familiarity
Showcasing AI Knowledge: Teachers who openly discuss their AI expertise discourage students from attempting to pass off AI work as their own.
Modelling AI Use: Incorporating AI into lessons (e.g. generating research questions or summarising readings) helps students see its value as a learning aid rather than a shortcut.
✅ Highlighting AI’s Inaccuracies
Demonstrating AI failures in real-time—such as incorrect summaries or flawed responses—teaches students to be skeptical and verify AI-generated content.
✅ Proving AI Writing is Detectable
Students are often unaware of how distinguishable AI-generated writing is. By having students compare AI-written and original paragraphs, they may quickly recognise the lack of authenticity in AI-generated text.
Lesson Takeaway: If students can spot AI writing, their teachers can too. This realisation discourages academic dishonesty.
Comments