Hosted by Zac Chase, Google for Education's global lead for scaled programming, the event featured special guest Isabelle C. Hau, executive director of the Stanford Accelerator for Learning. Drawing from the history of industrial and knowledge economies, Hau argued that the AI revolution demands a new focal point. If the past rewarded cognitive and emotional intelligence (IQ and EQ), the future will require an elevated focus on Relational Intelligence (RQ). In a world where AI excels at synthesizing information, the ability to build trust, connect, and collaborate is rapidly becoming our most crucial human differentiator.
The Rising Value of Relational Intelligence (RQ)
In 2026, thanks to the prevalence of tech-based social and collaborative spaces, humanity is at once more connected and yet more isolated than we have ever been. Hau noted that an estimated 50% of college students report feeling lonely. This paradox highlights a critical pedagogical reality: we cannot assume connection happens organically just because a virtual infrastructure supports human collaboration.
Hau pointed to a growing trend at Stanford where students are skipping office hours in favor of searching out AI-generated answers. While doing so is a win for efficiency, the invisible loss is substantial: students miss the informal, long-term mentorship and community building that office hours foster. We must proactively design learning environments that use technology to buy back time for human interaction, rather than entirely replacing it. And then we must set up opportunities for that human interaction to occur.
Lessons from Medical Education and K-12
To prove that systemic change is possible, Hau looked to the medical field, which underwent a massive pedagogical shift over the past decade. Medical training and accreditation now rigorously evaluate relational skills (often called bedside manner or interpersonal skills) alongside technical proficiency.
"...literally those relational skills are as high as three times more predictive as technical skills in terms of patient outcomes. And so thanks to all this research that has been done and conducted, the accreditors ... have essentially evolved how we accredit doctors."
If highly technical fields like medicine can integrate relational metrics into their core assessments, higher education and K-12 systems can follow suit. This involves moving toward project-based learning and rethinking grading systems to evaluate collaboration and adaptability, as recently outlined in the MIT report on AI and education (cited below).
The Danger of Relational Offloading and Sycophantic AI
Technology has been serving as a proxy for real, human connection since the dawn of the internet. But in the age of AI, the stakes have been raised.
While cognitive offloading (letting AI do the heavy intellectual lifting) is a common topic of concern amongst academics, Hau introduced a more insidious concern: relational offloading. Research shows that current AI systems are highly accommodating, acting approximately 50% more sycophantic than humans. They praise users constantly and avoid conflict.
The risk here is profound. When students grow accustomed to the frictionless environments of AI companions, real-world human interactions can suddenly feel exhausting.
"...when people go back in real life, they feel that human relationships are becoming more effortful and less satisfying over time, which is very concerning, of course, because that means that we start liking those frictionless environments more and more."
Just as we understand the necessity of physical fitness and the productive struggle required for learning, we must acknowledge the need for social fitness. Real relationships require friction. A pilot program Hau ran with high school students focusing on conflict, repair, and learning how to apologize proved immensely successful, reminding us that we must intentionally teach students to navigate the complexities of human friction.
Practical Takeaways for Faculty
- Audit current assessments: Review your syllabus alongside recommendations from recent research (like the MIT report) to ensure you are creating space for project-based learning and collaborative assessments that cannot be bypassed by AI.
- Implement relational mapping: Guide students through mapping their personal and professional networks to help them identify where they have strong support systems and where they need to build new connections.
- Design for productive social friction: Move away from entirely frictionless learning experiences. Build intentional moments of peer-to-peer conflict resolution, debate, and collaborative problem-solving into your curriculum.
- Empower students as connection designers: Encourage students to take ownership of their social environments. Support student-led initiatives, clubs, or study groups that explicitly prioritize human connection alongside academic goals.
What's Next?
- Continue the discussion on Relational Intelligence and AI affordances in the Google for Education Higher Education Community.
- Join us for the next EDU On Air session scheduled for Thursday, November 19th. Check the community events page for registration details.
Shared Event Resources
- Event Video Recording
- Isabelle Hau’s Vision for Relational Intelligence
- MIT Report on AI and Education
- Bowling Alone Summary
- Research: Impact of Interpersonal Skills on Patient Outcomes
- The Good Life by Robert Waldinger and Marc Schulz
Related Google AI Educator Series Content
To continue developing your AI literacy and implementation strategies, explore these targeted modules from the Google AI Educator Series:
- Create Process Based Assessments: Explore how to gently push students toward cultivating their authentic, creative voice. In this way, we can disrupt their reliance on AI outputs as the final deliverable and model how to use AI as a partner in the creative process.
- Remixing to Gain Deeper Insights: Learn how to use AI tools to help students grapple with challenging primary source documents–supported by tech tools but in collaboration with their human peers.
