Raising AI by Lowering Expectations

Key Takeaways

• Fear-based AI framing limits constructive safety solutions.
• De Kai misattributes parental responsibility to end users, not developers.
• Book contains unsupported claims, termed “neginformation.”
• Effective AI governance requires scientists and engineers as primary “parents.”
• Calls for user PTAs lack practical impact and misdirect effort.

Pulse Analysis

The language we use to discuss artificial intelligence influences how stakeholders perceive risk and opportunity. Decades of science‑fiction narratives have conditioned many to view AI as a looming threat, prompting safety researchers to focus on defensive tactics like red‑team testing and jailbreak prevention. Reframing AI as a nascent entity that can be nurtured opens space for collaborative governance, encouraging developers to embed ethical values early in the design process rather than relying solely on post‑deployment controls. This shift aligns with emerging AI governance frameworks that prioritize transparency, alignment, and shared stewardship over punitive regulation.

De Kai’s *Raising AI* captures the appeal of this nurturing metaphor but misplaces the locus of responsibility. By casting ordinary users as the primary “parents,” the book overlooks the fact that model behavior is shaped long before content reaches an individual’s feed. Scientists design architectures, engineers curate training data, and product teams set deployment parameters—roles analogous to genetic, prenatal, and early‑childhood parenting. The author’s reliance on unverified anecdotes, labeled “neginformation,” further erodes credibility, risking the spread of misconceptions that could distract from evidence‑based safety research and policy development.

For AI governance to be effective, the parenting metaphor must be applied to those who actually build and control the technology. Policymakers should focus on mandating robust development standards, auditability, and accountability mechanisms for creators, while industry can foster internal “AI ethics committees” that act as parental figures overseeing value alignment. Meanwhile, end‑users can contribute through informed engagement—diverse interaction, feedback loops, and digital literacy—but their role is better described as stewardship rather than parenthood. Aligning framing with the true custodians of AI will promote actionable solutions, reduce learned helplessness, and accelerate the development of safe, beneficial systems.