Sam Altman’s 2025 TED Talk: A Vision for AI and Agentic Engineering with Police

Exploring the Future of AI in Governance, Policing, and Procedural Justice

Sam Altman, a prominent figure in the field of artificial intelligence and innovation and the ChatGPT CEO, made waves during his live TED Talk interview in 2025. His discussion revolved around critical and transformative topics, including agentic engineering and the profound implications of AI learning not only from societal structures but also from individual opinions worldwide. As he explored these themes, Altman delved into both the opportunities and challenges of this approach, particularly in the context of governance, and collective decision-making. This article unpacks the key insights from his talk and examines their potential to reshape societal trust and justice with respect to policing.

Agentic Engineering: A New Frontier in AI

Agentic engineering, as Altman described, represents a paradigm shift in artificial intelligence. Unlike systems that merely automate tasks or mimic human cognition, agentic engineering seeks to imbue AI with a sense of agency—enabling it to act autonomously while remaining responsive to human needs. It appears that this technology’s core principle lies in enhancing AI's ability to learn dynamically from society as a whole, transcending geographic, cultural, and economic boundaries.

In practice, this approach ensures AI doesn’t merely draw from the perspectives of elites or centralized power structures. Instead, it gathers input from diverse populations, giving voice to individuals who have historically been underrepresented. Altman argued that such inclusivity could democratize AI decision-making in ways never before seen, paving the way for systems that truly reflect the complexities and aspirations of humanity.

Learning from Society and Amplifying Individual Voices

I suspect a pivotal aspect of Altman’s vision is the ability of AI to learn from society in a holistic manner. Imagine how AI, equipped with agentic capabilities, could tap into the collective wisdom of millions across the globe, actively seeking out individual opinions rather than relying on a narrow subset of decision-makers. This would mitigate the risks of elitism and ensure that diverse perspectives, including those from marginalized communities, are integrated into AI’s decision-making processes.

I would assert that the transformative potential of such systems in governance is tremendous. By synthesizing feedback from vast and varied populations, AI could provide police leaders with a clearer understanding of public sentiment and needs. Sam suggested collective governance decisions are to come. It seems there are scenarios where collective governance decisions, supported by AI, might unite fractured societies and empower people to make wiser choices, potentially reducing societal issues like crime and inequality.

Policing and Procedural Justice: Risks and Rewards

One of the most compelling sections of Altman’s talk were the implications of agentic engineering in policing. I certainly agree while the benefits of gathering true feedback from underrepresented populations are likely evident, I also acknowledge the risks posed by losing the human-in-the-loop element with respect to policing. When AI systems operate with a high degree of autonomy, especially in sensitive areas like public safety, the absence of human oversight could lead to unintended consequences, or will this be yet another disruption where it becomes impossible for human-in-the-loop?

Challenges in Policing Without Human Oversight

I would caution against the potential dangers of relying exclusively on agentic AI in policing. Without human input to contextualize decisions, there’s a risk of actions being perceived as cold, detached, or overly algorithmic. For instance, an AI system might make policing decisions based solely on statistical analyses, inadvertently reinforcing biases or implementing measures that lack empathy. Such outcomes could erode public trust and provoke resistance from communities already skeptical of police.

Moreover, I would point out the ethical dilemmas of delegating life-and-death decisions to AI. I must stress the need to strike a balance between leveraging AI's efficiency and maintaining the humanity that underpins procedural justice.

Building Trust Through Inclusive Feedback

Despite these challenges, I see on the horizon the profound opportunities agentic engineering offers for improving policing. By gathering and analyzing feedback from diverse populations, AI could help public safety agencies better understand the needs and concerns of the communities they serve. This approach could lead to measures that genuinely address societal grievances—reducing police overreach, fostering accountability, and strengthening community relationships.

I would argue that such systems might represent the next step in procedural justice. By listening to underrepresented voices and incorporating their perspectives into policing strategies, AI could help build a foundation of trust and fairness. This, in turn, might enable police to move away from punitive measures and toward a model that prioritizes rehabilitation, prevention, and restorative justice.

Collective Governance and Crime Reduction

Beyond policing, there are broader implications of collective governance decisions supported by agentic AI. Like Altman I envision a future where AI facilitates global cooperation, helping nations and communities align on shared goals. This could be particularly impactful in addressing systemic issues like crime.

Altman suggested that AI’s ability to synthesize feedback from diverse populations might lead to wiser policy decisions— and I suggest, these are ones that tackle root causes of crime, such as poverty, inequality, and lack of education. By fostering dialogue and collaboration across societal divides, collective governance could help reduce crime not through enforcement alone but by addressing the conditions that give rise to it.

The Path Forward

Sam Altman’s TED Talk in 2025 offered a glimpse into a future where agentic engineering transforms AI from a tool into a partner in shaping society. While the challenges of implementing such systems are significant, the potential rewards—more inclusive governance, fairer policing, and reduced crime—are equally compelling. Altman’s vision invites us to consider whether the procedural justice of tomorrow lies not in the hands of the few but in the collective voices of humanity.

As we look ahead, the key question remains: Can AI truly unite us and help build a more just and equitable world? Altman’s insights suggest that the answer may depend on our ability to balance technological innovation with ethical integrity, ensuring that the AI we create reflects the best of who we are.

TED (2025) OpenAI’s Sam Altman Talks ChatGPT, AI Agents and Superintelligence — Live at TED2025. Retrieved from youtubecom/watch?v=5MWT_doo68k&t=1s