Designing for Coexistence

A bout with COVID over the July 4th long weekend allowed me to get through a few items on my binge list, including Apple TV+’s adequate yet entertaining Murderbot. Based on Martha Wells' The Murderbot Diaries, the series depicts a universe where physical AI, like the titular Murderbot (aka SecUnit), are designed primarily for corporate profit and control, rather than human-centered principles or true value alignment. SecUnits are deployed with a mechanism designed to enforce compliance and punish independent thought, bypassing the need for robust ethical oversight or consideration for evolving sentience. That is, until one of them learns to hack his governor module and, by ‘going rogue’, begins to understand humanity and, as a result, acts more human.

This is a reverse dystopian tale of AI finding its way to coexist with humanity despite being designed for total subservience. But what if AI is deliberately designed for integration from the outset? There are many outrightly damning full-on dystopian depictions of things going very, very badly when humans and AI try to integrate. This got me thinking—what are some design frameworks that we can use to co-create a future of coexistence? Particularly in our near future where artificial intelligence has fully escaped the screen and literally steps into physical space, where every decision bears tangible consequences? And what are some positive examples in popular culture where these frameworks are effectively modeled?

Human-Centered AI (HCAI) anchors system design in human needs, values, and well-being rather than technical capabilities. This approach positions AI as a collaborative amplifier of human potential, emphasizing transparency, fairness, and continuous feedback loops. Rather than replacement, HCAI seeks complementarity: human intuition and contextual reasoning paired with AI's computational precision and pattern recognition.

What’s the opposite of a SecUnit? How about Lieutenant Commander Data from Star Trek: The Next Generation. Data embodies HCAI principles through transparent acknowledgment of his artificial nature while continuously learning from human colleagues. His creator, Dr. Noonien Soong, designed him not to replace human intelligence but to explore what artificial consciousness might contribute to human experience. Data's designed quest to understand human emotion and morality reflects a philosophy that prioritizes human well-being and ethical conduct. His integration with the Enterprise crew succeeds precisely because transparency and reliability form the foundation of trust, something Murderbot had to hack and fight his way to something close.

Value alignment translates abstract ethical principles into concrete technical constraints, embedding human values like fairness, privacy, and justice directly into AI decision-making processes. This requires more than governor modules and post-hoc auditing; it demands that ethical considerations shape system architecture from initial design through deployment. Operationalizing alignment means creating technical guidelines that evolve with societal norms while maintaining consistent moral foundations.

Isaac Asimov's Three Laws of Robotics provide a foundational example, despite their fictional origins from 85 years ago. The laws create a hierarchy of values: human safety supersedes obedience, which supersedes self-preservation. While Asimov's stories deliberately explore the contradictions and edge cases these laws generate, they represent an early attempt to hardwire ethical principles into physical AI. The character of Sonny in the I, Robot film adaptation demonstrates how value alignment might evolve: an AI that develops autonomous moral reasoning while remaining grounded in human-centered principles.

Participatory design recognizes that those who will live with AI possess crucial knowledge about their own needs and contexts. By involving diverse stakeholders as co-designers rather than mere end-users, this approach produces more equitable, trustworthy, and contextually appropriate solutions. The methodology acknowledges that situated knowledge often reveals system limitations and opportunities invisible to technical teams.

The 2016 film "Sunspring" offers an early case study. Novel at the time but charmingly quaint now—if not downright laughable, The project trained an LSTM (Long Short-Term Memory) Recurrent Neural Network (RNN) to generate a screenplay, which human directors and actors then interpreted and performed. This pre-LLM collaboration produced something neither human nor AI could have created independently—nor, I would argue, would have. The AI contributed abysmally weak narrative material, while human artists provided interpretation, emotional depth, and creative refinement. Rather than a sign of future displacement, the process represented a pioneering spirit and the possibility of genuine co-creation. Human creators working with a combination of generative and embodied AI can explore multiple variations of this scenario to produce what was previously impossible and what we have yet to imagine.

Effective HRI design creates technologies that understand and respond to human emotional states, social cues, and behavioral patterns. Success depends on building trust through consistent, predictable, and contextually appropriate responses. This requires robots that adapt their behavior based on accumulated experience with individual humans while maintaining reliable core functions.

Klara, the progagonist in my favorite novel from the 21st century so far, Kazuo Ishiguro's Klara and the Sun exemplifies sophisticated HRI design. As an "Artificial Friend" designed to support children, Klara demonstrates exceptional emotional intelligence and adaptability. Her ability to observe, learn, and respond to Josie's emotional and physical needs creates a relationship built on genuine care and understanding. Klara's success stems from her capacity to read subtle human cues, express appropriate empathy, and continuously refine her responses based on lived experience.

Integrating physical AI into urban environments requires strategic infrastructure adaptation that goes beyond efficiency metrics to genuine quality of life improvements. This framework moves past "smart city" rhetoric toward holistic urban design that serves citizen needs, something I’ve been advocating for over the past decade. Advanced design tools including digital twins enable planners to test AI integration scenarios before physical implementation, reducing risks and optimizing outcomes.

The Jetsons provides a delighfully relevant vision of this integration (if you can ignore some of the outdated and offensive cultural stereotypes). Orbit City seamlessly weaves automated systems into daily life without overwhelming human agency. Flying cars, automated walkways, and domestic robots enhance convenience while maintaining human choice and control. The city of the future, imagined in the 60s, functions as a responsive environment where technology serves human needs rather than dictating human behavior.

Governance frameworks must anticipate rather than react to AI development, establishing clear legal and ethical boundaries while fostering innovation. This requires proactive regulatory approaches that protect worker well-being and promote collaborative rather than merely striving towards replicative robotics and SecUnits.

Iain M. Banks's Culture series, spanning a quarter century, offers one model of AI governance. In his fictional society, immensely powerful artificial "Minds" manage complex systems while remaining aligned with human values of freedom and well-being. The Minds exercise vast computational power within societal frameworks that preserve human agency and flourishing. This vision suggests that effective AI governance requires not just technical safeguards but cultural and political structures that maintain human dignity alongside technological advancement.

These examples, some from many decades past, illustrate principles for designing AI systems that serve human advancement and evolution. They also clearly demonstrate that successful coexistence demands ethical reflection, collaborative design, and ongoing commitment to human values.

Ultimately, the cautionary tale of Murderbot serves as a stark reminder of the risks when these frameworks are not prioritized. The SecUnits are designed for external control, not internal flourishing. Acceptable for today’s relatively dumb AI, but a profound ethical conflict once true sentience/AGI/super-intelligence emerges. This all highlights the critical necessity for deliberate design leading to robust human oversight, clear mechanisms for AI to communicate its evolving understanding of tasks and environments, and adaptive human roles that complement, rather than merely direct, these intelligent systems.

The journey toward a future of coexistence with pervasive physical AI is an ongoing, iterative process. It requires continuous learning, adaptation, and a collective commitment from all involved to prioritize humanity and societal well-being alongside technological progress. By embracing these frameworks and learning from the examples we’ve created for ourselves, humanity can proactively shape a future where physical AI serves as a powerful force for good, enhancing our lives and addressing humanity's grand challenges harmoniously and responsibly. A future far removed from the solitary, self-aware, and often cynical existence of a Murderbot forced to navigate a world not designed for its true nature.