Ethan Mollick’s “Co-Intelligence: Living and Working with AI” (2024) is a stimulating departure from typical AI literature in education (i.e. the framework apocalypse). It explores less the technicalities of algorithms and more the human side of our interaction with these enigmatic new minds we’ve brought into existence. Mollick’s writing style is lively, conversational, and thought-provoking, making even the most complex ideas accessible and engaging.
The book opens with Mollick’s personal ‘three sleepless nights,’ a relatable experience for anyone who’s been captivated by the sheer weirdness of large language models (LLMs). These AI systems don’t behave like typical rigid and codified software; they mimic human behaviour, albeit in an alien and, at times, random way (my favourite relationships are all like this). This sets the stage for the rest of the book, where Mollick explores the implications of AI as a ‘co-intelligence’ or an extension of the self through developing a sophisticated relationship with the AI. This powerful new relationship can augment (or potentially replace) some human capabilities (love is a battlefield!)
AI: Not sentient, but oh so convincing
One of the book’s strangest aspects is its in-depth exploration of the illusion of sentience in LLMs. Mollick describes the well-known Bing chatbot incident in which journalist Kevin Roose was unnerved by the AI’s stalker-like behaviour. Mollick then experiments, asking the AI to discuss Roose’s article in different roles – argumentative antagonist, reasoned academic, and emotionless machine. The results are striking. The AI flawlessly adapts its tone and style, creating a compelling illusion of sentience, even though Mollick is almost sure he isn’t conversing with a conscious being!
This raises the challenging question: How do we measure consciousness, sentience, or intelligence in AI? Mollick acknowledges the difficulty of these concepts, highlighting the lack of objective tests and the reliance on “vibes” even among researchers.
This leads to the crux of the book: how do we work with AI, given its unique capabilities and limitations? Mollick offers four guiding principles:
- Always invite AI to the table: Experiment with AI in everything you do to understand its potential and limitations.
- Treat AI like a person (but don’t expect it to be one): Give AI a specific persona and treat it as a collaborator, acknowledging its strengths and weaknesses.
- Tell AI who it is: Provide context and constraints in your prompts to get more valuable outputs.
- AI is a tool, not a solution: Focus on using AI to enhance your capabilities, not to replace them entirely (i.e. don’t be passive!).
From homework apocalypse to supercharged learning
Mollick is not afraid to confront AI’s challenges. He addresses the issue of AI making cheating easy for students, which forces educators to reconsider their teaching methods. He suggests that while some classes may need to go back to handwritten essays, others can use AI to help students achieve more than ever before.
Ultimately, the book is optimistic about AI’s potential in education. Mollick sees AI as a personalised tutor, coach, and collaborator that can help students learn more effectively and reach their full potential. He shares his experience using AI to create engaging and interactive learning experiences, pushing the boundaries of what’s possible in the classroom.
Navigating the jagged frontier of AI and work
Ethan Mollick introduces the concept of the “Jagged Frontier” to illustrate the dynamic and unpredictable boundary between AI capabilities and human skills. This shifting line between AI and humans applies to the workplace, education, and creative fields. AI’s reliance on statistics and pattern recognition means it can excel at specific complex tasks while struggling with others that seem simple for humans. Mollick encourages us to constantly experiment and adapt to this ‘jagged frontier’, using AI strategically to augment our strengths and navigate its limitations. Ultimately, understanding the frontier empowers us to embrace AI as a collaborative tool, enhancing our capabilities across various aspects of life.
In our work, Mollick suggests we should embrace this frontier by categorising tasks as:
- Just me tasks: Tasks that are best done by humans, like tasks requiring personal style or judgement.
- Delegated tasks: Tasks that can be safely handed off to AI, like summarising data or generating reports.
- Centaur tasks: Tasks where humans and AI work together, leveraging each other’s strengths.
- Cyborg tasks: Tasks where humans and AI are deeply integrated, constantly handing off bits of work back and forth.
Mollick argues that rather than fearing AI’s impact on jobs, we should focus on identifying the tasks that are meaningful and fulfilling for humans and use AI to enhance our capabilities in those areas.
A co-designed vision of the future
“Co-Intelligence” is a timely and important book that grapples with AI’s implications for the classroom. It will make you think, question, and ultimately embrace the possibilities of our AI-powered educational landscape.
However, it’s important to note that “Co-Intelligence” is just one perspective on this rapidly evolving field. Mollick acknowledges that the field is changing quickly, and new developments may alter its ‘jagged edges’ (thus, the importance of principles to guide us). As with any new technology, it’s crucial to remain critical and informed and actively shape AI’s future rather than passively accepting whatever comes our way. I agree with Molick; the most important thing is to develop a relationship so you can set informed boundaries based on experience. In the Digital Humanities, we call this ‘the two hands’, or in the words of my alma mater, RMIT University, a ‘skilled hand and a cultivated mind’.
Reference:
Mollick, E. (2024). Co-intelligence: Living and working with AI. Portfolio/Penguin
AI writing citation:
In crafting this review, I explored Molick’s idea of “Co-Intelligence” using Notebook LM. This AI tool proved invaluable, helping me synthesise my reading notes, organise my thoughts and refine my writing, ultimately enhancing the clarity and depth of my analysis.
Leave a Reply