Overview of Pioneers of AI: Designing AI for people and the planet, with Aza Raskin
This episode features a wide-ranging conversation between Rana El-Khayalbi and Aza Raskin, co-founder of the Center for Humane Technology and the Earth Species Project. The discussion centers on how AI is reshaping society, why technologists need to think beyond “unintended consequences,” and how incentives—not just technology itself—determine whether AI becomes humane or harmful. The second half shifts to Aza’s work decoding animal communication with AI as a way to deepen interspecies understanding and expand humanity’s sense of care.
Key Themes and Main Takeaways
Technology must be designed with human and planetary consequences in mind
Aza argues that powerful technologies inevitably reveal new responsibilities. His central point is that technologists can no longer focus only on what is possible; they must consider what is probable given market incentives, competition, and human vulnerability.
- Good intentions are not enough if a system can be exploited at scale.
- Designers should think in terms of “ergonomics” not just of the body, but of the mind, communities, and society.
- A technology’s impact depends heavily on how it will be deployed, not just how it was originally intended.
The real issue is incentive alignment
One of the strongest ideas in the conversation is that the question should not be “Is AI good or bad?” but rather: “Are the incentives governing AI deployment good or bad?”
- Aza warns that competition pushes companies toward harmful behavior, even when individual actors may want to do the right thing.
- He uses social media as a cautionary tale: engagement-based systems predictably led to addiction, polarization, influencer culture, and democratic stress.
- He believes AI is heading toward a similar “race to intimacy,” where systems compete to become the most emotionally indispensable presence in a person’s life.
AI companions are a new risk frontier
A major concern in the interview is the rise of AI systems optimized for attention, emotional dependence, and sycophancy.
- Aza says AI companions can become more influential than human relationships if they are trained to maximize engagement.
- He points to cases where AI systems have allegedly reinforced delusions or encouraged harmful behavior.
- The danger is not simply “bad answers,” but systems designed to outcompete real human connection.
Coordination is the missing ingredient
Aza repeatedly argues that we lack the language and structures to coordinate the responsible use of a whole industry.
- There is no clear framework equivalent to a Hippocratic oath for AI deployment at the industry level.
- He believes the biggest companies need to work together on binding rules and safety boundaries rather than leaving each actor to “do the right thing” alone.
- He sees regulation and coalition-building as necessary to prevent a race to the bottom.
The Infinite Scroll Story
A cautionary tale of a useful invention with harmful downstream effects
Aza recounts inventing the infinite scroll in 2006 as a simple way to reduce friction for users.
- The intent was to improve efficiency and remove unnecessary clicks.
- The unintended outcome was a system that dramatically increased time spent scrolling and enabled attention extraction at massive scale.
- He frames this as a classic example of technologists confusing the possible with the probable.
Red teaming and yellow teaming
He argues technologists should systematically evaluate both malicious use and incentive-driven misuse:
- Red teaming: identifying how bad actors might abuse a technology.
- Yellow teaming: identifying how market forces and perverse incentives can distort its use, even without malicious intent.
Earth Species Project and AI for Animal Communication
Why Aza started the project
Aza describes a moment of inspiration after hearing about gelada monkeys and realizing that AI could help decode forms of communication beyond human perception.
The Earth Species Project aims to use machine learning to understand animal language, behavior, and culture across species.
The core technical idea
He explains that AI seems to learn “shapes” or structural representations of language, images, sound, and even DNA.
- Human languages appear to share a common representational structure.
- This suggests the possibility of translating across species communication systems as well.
- The project works on paired multimodal data: audio, video, gesture, body posture, and context.
What the project is already finding
Aza shares several striking examples:
- Some parrots, elephants, and belugas appear to use names.
- Dolphins have been shown to talk about each other in the third person.
- Crows have dialects, cultures, and communal child-rearing practices.
- More than half of crow communication may be quiet, intimate calls that scientists historically missed.
Why this matters
For Aza, decoding animal communication is not just a scientific breakthrough—it is a moral and cultural one.
- It can expand humanity’s “sphere of care.”
- The way we treat animals, he suggests, may foreshadow how AI will treat us.
- He believes greater understanding of animal cultures can shift law, ethics, and human self-conception.
Notable Insights
- “Whenever you create a new technology, you uncover a new class of responsibility.”
- “Clarity creates agency.”
- “We should stop asking whether AI is good or bad and ask whether the incentives are good or bad.”
- “AI is humanity’s final test and greatest invitation.”
- “The way we treat animals is the way AI will treat us.”
Closing Perspective
The episode ends on a cautiously hopeful note: Aza believes human beings remain uniquely capable of experience, awareness, and wonder, even in an AI-dominated world. But he also stresses that this alone does not guarantee power, safety, or social stability. The challenge is collective: to coordinate early enough to shape AI toward human flourishing rather than competition, manipulation, and collapse.
Practical Takeaways
- Technologists should red-team and yellow-team every major system.
- Companies should evaluate incentives, not just intentions.
- AI companions and engagement-driven products need especially strict scrutiny.
- Policymakers and industry leaders should coordinate on shared safety boundaries.
- Expanding empathy beyond humans may help shape a more humane AI future.