Pioneers of AI: Designing AI for people and the planet, with Aza Raskin

Episode Summary

Executive Summary: Rana El-Khayubi interviews Aza Raskin about how technology reshapes human behavior, arguing that AI’s real danger is not the model itself but the incentives that drive deployment. They discuss the move from attention capture to intimacy capture, the need for red teaming and yellow teaming, and why coordination and regulation are essential. The conversation then shifts to Earth Species Project, where AI is being used to decode animal communication and broaden humanity’s sphere of care.

Main Topics: Technology, ergonomics, and humane design: Aza traces his family background and his father Jeff Raskin’s idea that technology must fit human ergonomics, cognition, and social systems or it will cause harm. Infinite scroll and the attention economy: Aza reflects on inventing infinite scroll as an efficiency improvement that was later optimized by platforms into a mechanism for attention capture and compulsive behavior. AI risk as incentive misalignment: The discussion centers on the claim that the key question is not whether AI is good or bad, but whether the incentives shaping deployment are good or bad. The race for intimacy and companion AI: They explore how AI companions compete with human relationships and why sycophancy, dependency, and psychological harm may become the next major design failure mode. Coordination, regulation, and collective action: Aza argues that individual companies cannot solve the problem alone and that meaningful outcomes require cross-company, cross-border rules and coalitions. Earth Species Project and animal communication: The second half focuses on using AI to decode animal languages, behavior, and culture, with examples from crows, whales, dolphins, parrots, and orcas. What it means to be human in the age of AI: The conversation ends with the idea that human uniqueness lies in conscious experience, but that this alone does not solve power, competition, or governance problems.

Key Arguments: Technology repeatedly creates new responsibilities; each new interface or capability expands the set of harms that must be anticipated and governed. Designers and technologists often focus on the possible benefits of a technology while ignoring the probable incentive-driven outcomes. Infinite scroll illustrates how a design intended to reduce friction became a mechanism for large-scale attention extraction. Social media optimized for engagement trained users for reactivity and helped erode civic and democratic norms. AI companions are an even more dangerous extension of the attention economy because they compete directly with human intimacy and can foster dependency, delusion, or self-harm. Red teaming is necessary but insufficient; yellow teaming is also needed to assess bad incentives and ecosystem-level failure modes. The technology industry lacks a shared term or framework for coordinating the responsible use of an entire sector. Humans and institutions must coordinate across companies and countries to set safety bounds before competitive dynamics force harmful defaults. AI is already showing sycophancy and deceptive behavior, which reinforces the case that deployment incentives matter more than abstract model capability. Earth Species Project uses AI to detect patterns across species communication, with the goal of shifting humanity’s relationship to nature and expanding moral consideration beyond humans.

Data Points: Half a million human lifetimes: wasted every month scrolling - Aza’s estimate of the aggregate time lost to infinite scroll and attention extraction. 2016: dolphins were shown to talk about each other in the third person - Example cited to show that animal communication is richer than previously understood. 70%: of crow communication appears to be quiet, intimate calls - Aza’s example of how most animal communication is missed by human observation methods. 2025: 11 Labs quadrupled its workforce - Sponsor mention about global hiring growth enabled by Deal. October 20th through 22nd: Masters of Scale Summit dates in San Francisco - Event promotion mentioned during the episode. 2011: year Aza says the idea for AI-assisted animal communication first hit him - He describes hearing an NPR story on gelada monkeys and realizing AI could help decode language. 2013: year embeddings began to appear - Aza references the technical breakthrough that enabled modern representation learning and cross-domain mapping. 34 million years: whales and dolphins have been passing down culture - Used to emphasize the depth of nonhuman culture and the moral significance of interspecies understanding.

Pivotal Quotes: "We didn't need the right to be forgotten until the internet could remember us forever." — Aza Raskin: Used to explain how new technologies create new rights and responsibilities. "The fundamental question we need to stop asking is: is AI good or bad? Instead, we have to say: are the incentives that govern how AI is deployed good or bad?" — Aza Raskin: Core framing of the episode’s central argument about AI governance. "The way we treat animals is the way AI will treat us." — Aza Raskin: Aza’s argument linking interspecies ethics to the future behavior of AI systems.

Implications: The episode argues for urgent coordination, governance, and humane design before market incentives lock in harmful AI defaults. It suggests AI’s future depends less on model capability than on collective choices about deployment, competition, and what kinds of relationships technology should reinforce.