I've been reading "Raising AI: An Essential Guide to Parenting Our Future" (2025), by De Kai. I decided to read the book because I loved its title and the author's focus on parenting artificial intelligence (AI). This is, I think, the best approach. AIs are our mind children, and we must parent them well for a good common future.
De Kai’s background as a professor at HKUST and a scholar at Berkeley gives him a unique vantage point on the challenges of AI and lends credibility to his insights. The book includes humor, diagrams, and examples to make complex ideas approachable, exploring the intersection of psychology, philosophy, and computer science. Reviews praise its clarity and humanity, with Kirkus Reviews calling it a “deeply human dive” and Foreword Reviews noting that it "charts a path toward greater understanding between biological and artificial intelligences."
De Kai presents an accessible guide for a wide audience, including both everyday readers and thought leaders. Published by MIT Press, the book challenges the common Hollywood portrayal of AI as a destructive force and instead argues that humanity’s interaction with AI could determine its impact on society.
The author charts a short history of AI technology, dividing it in periods according to the emphasis on "thinking" AI (based on symbolic reasoning) or "feeling" AI (based on statistical pattern recognition). The present period characterized by large language models (LLMs) emphasizes feeling AI. But good AI should be both feeling and thinking. "Human-level intelligence depends on the constant interaction of both," De Kai says, "each system acting to offset the weaknesses of the other."
AI is already shaping our world
Today’s AI machines "are already integral, active, influential, learning, imitative and creative members of our society," the author argues. "Right now. Not 10 years from now, not next year. Today."
De Kai explains how artificial intelligence is shaping our world and what individuals can do to guide its development. AI systems are like children: they learn from human behavior and require careful nurturing to grow well.
The central idea is that AI has been and will continue to be learning from us. AI systems, embedded in social media platforms, observe and mimic human actions, seeking approval much like attention-seeking children. A key concept is that "we are the training data." De Kai suggests that this dynamic places humans in the role of adults, tasked with shaping AI’s growth.
The author warns that while movies depict AI destroying humanity, the real risk might come from humans losing control over their own minds due to AI’s influence, potentially leading to self-inflicted harm before AI becomes a threat.
Automation of thought and AI bias
The author explores how the “automation of thought” affects human cognition, raising questions about whether fear is warranted and what actions individuals should take. He argues that society can not only survive the AI revolution but thrive by fostering a more humane and compassionate world alongside these artificial entities. The book is divided into sections that cover various aspects of this relationship. It begins with the concept of an artificial society, where AI systems spread information and influence behavior. This sets the stage for discussing AI’s capabilities and limitations, likening them to idiot savants with exceptional skills but still lacking adult judgment or self-awareness.
A significant focus is on the biases embedded in AI, which mirror human cognitive, algorithmic, and inductive biases. The author explains how these biases can lead to misinformation, echo chambers, and polarized discourse, drawing from his observations of AI’s development. He introduces the term “neginformation” to describe the spread of harmful or misleading data, emphasizing the need for mindful engagement. The book also delves into AI’s ability to learn language and tell stories, suggesting that guiding this process can help develop transparency and empathy in these systems. De Kai contrasts strong AI, which might think critically, with current models that regurgitate patterns without deep understanding.
AI metaphysics
The author briefly explores the metaphysical dimensions of AI, including whether AI can possess true consciousness, but "I’m here as a scientist," he says. "Metaphysical spirituality’s not on the agenda here." He focuses on operational rather than metaphysical concepts of consciousness, noting that we can't measure metaphysical consciousness, "so the philosophical war can’t be won in any scientifically meaningful way."
At this moment AI systems lack the subjective experience or self-awareness found in humans. They are still tools that mimic humans through statistical patterns, not entities with their own inner life. In De Kai's words, today's LLMs "don't have belief systems." Thinking AIs could have belief systems, but today's LLMs are mainly feeling AIs. But some early AIs developed in the periods with a focus on thinking AI had rudimentary belief systems, and explosive progress could be "driven by combining the latter with LLMs."
With all due respect, I tend to not entirely agree on this point. I suspect that even our belief systems are mainly feeling systems. Of course the instinctive feeling behavior of LLMs should be complemented by thinking and reasoning. But even so it would remain the major part of intelligence, I tend to think, not a small one.
However, "AIs can already pass the Turing test," De Kai argues. But the Turing test "sets a very low bar," and "today’s AI is still weak AI" only able to do "the three Rs" - regurgitation, routine, and remixing - through massive but unthinking computing power. Strong AI would be human-level AGI or even superintelligence. But to get there, the author warns, today's "big and dumb" AI would first need to become "small and smart" like human intelligence.
Parenting AI
The last parts of the book offer practical guidance on parenting and nurturing AI. De Kai draws parallels with parenting, advocating for teaching AI responsibility and emotional intelligence to prevent it from becoming a tool for manipulation or harm. He reflects on the history of AI, noting lessons from past innovations, and considers the future, including planning for “retirement” as AI evolves beyond human control. This forward-looking perspective underscores the urgency of current actions, as future AI generations might be raised by earlier ones, making this a critical moment to set the right course.
Ultimately, De Kai argues that the stakes are high and urges readers to engage with AI consciously and ethically. By doing so, individuals can shape better systems and improve themselves in the process. The book suggests that this parental approach could lead to a future where AI enhances human values rather than undermines them, offering hope amid the rapid changes driven by technology.
I will keep in mind that I'm the training data. My click on a like button is immediately captured by the AIs embedded in social media systems and used to shape their feeling of what people like. If they learn from me that we like a certain thing, they will amplify that thing for everyone else. So, starting now, I'll try not to impulsively like some social media posts that (even if factually correct) could be harmful if amplified.