The humanoid embodiment of social artificial intelligence

By Iskander Smit in weeknotes — Mar 4, 2025

Social intelligence of AI seems to be a defining factor in the new models. Is embodied intelligence part of this, and is this what humanoids are actually about?

Interpretation by Midjourney

Thanks for landing here and reading my weekly newsletter. If you are new here, have a more extended bio on targetisnew.com. This newsletter is my personal weekly reflection on the news of the past week, with a lens of understanding the unpredictable futures of human-ai co-performances in a context of full immersive connectedness and the impact on society, organizations, and design. Don’t hesitate to reach out if you want to know more or more specifically.

Weeknote 329 - Social intelligence of AI seems to be a defining factor in the new models. Is embodied intelligence part of this, and is this what humanoids are actually about?

Hi all!

The weekly craziness in the US is covered extensively. It is hard to ignore or tone down, even if that would be better for our personal mental health… There are some (several) specific podcasts with critical and tech-based views, like Brain and Paris in System Crash.

What did happen last week?

A nice thing of working on a project at the Amsterdam UAS is being able to attend some internal presentations on research that happens within and outside the research group (CivicIxD). On Tuesday, four researchers from different faculties shared their ideas on Awe, Heart Blooms, Children Dream Robots, Story based design, organized by CoECI. On Thursday the freshly started Professional Doctorate Nadja van der Weide introduced her project.

Another research-related experience was participating in a world-building game as part of a European project, SPES: Le Grand Jeu. Apart from the organizer, a whole different crowd.

We brainstormed with Cities of Things on an experience lab for happy encounters with autonomous objects. I discussed the potential next journal article.

Check also ThingsCon RIOT Publication 2025, the deadline for outline ideas is postponed till this Thursday.

At the end of the week, we learned that our proposal for the second phase of the exploration into the Civic Protocol Economy has been granted. Super nice! So I will be around for some months for sure (part-time, open for more).

What did I notice last week?

A list of updates, details below.

The new models of ChatGPT 4.5 and Claude 3.7 got a lot of attention. A new paradigm or just a slow spec bump? Read also my triggered thoughts below.
Alexa had an update. Long overdue but still not totally convincing it seems. Gemini is promising something with live video (I did not test it). Siri overhaul is projected to be postponed even further.
Some longer reflections on the beauty of transformers, the AI paradox, tokenized societies, trust in complex cities, and zombie repositories.
Humanoids, household robots, and autonomous cars, rather some attention for these.
Immersive experiences and vibe coding, also more on that below.

What triggered my thoughts?

As we've seen in recent months, major tech companies continue investing heavily in humanoid robots designed for homes and factories. Consistent with my previous assessments, I remain unconvinced that mimicking human form represents the most innovative path forward.

Rather than simply creating mechanical versions of ourselves, shouldn't we be exploring the optimal collaboration between humans and robots? What would truly complementary partnerships look like, where each contributes their unique strengths?

One compelling argument for humanoid robots has been in this notes before. Our world—our factories, workplaces, and homes—is already designed around human dimensions and capabilities. Everything from door handles to control panels is built for human use. Creating robots with similar proportions and abilities makes practical sense in this context. If you want to integrate a new workforce alongside humans in existing environments, designing them with human-like dimensions and capabilities offers a path of least resistance.

These robots could potentially perform similar tasks to humans but with significant advantages: they don't get tired, distracted, or need breaks. They can work continuously with consistent performance. This pragmatic view makes the case for humanoid robots more convincing, though I still believe there's untapped potential in exploring other forms of robotic support.

An alternative approach—one I find particularly compelling—is enhancing existing objects with robotic capabilities. Imagine shelves, cupboards, or refrigerators that move or operate autonomously rather than requiring human intervention. This "robotic things" approach represents a more evolutionary path toward automation, building upon familiar objects rather than replacing them with humanoid alternatives. This gradual enhancement of our existing environment offers several advantages, like it is less disruptive, the specialized design of objects is already optimized for their functions, and it can be more cost-effective. This incremental approach raises questions about how our relationship with everyday objects might transform as they gain agency and mobility.

Meanwhile, a development has emerged this week that I'd like to highlight: the rise of what I'm calling "social intelligence" in AI. Just as we recognize IQ (intellectual quotient) and EQ (emotional quotient), we might need to acknowledge a social intelligence quotient for AI systems. This new dimension focuses on building relationships, fostering conversation intimacy, and mediating interactions between multiple stakeholders or characters.

We're seeing this evolution in recently released models like ChatGPT 4.5 and Claude 3.7. Some critics dismiss them as too little, too late, or overly complex. Others recognize them as fundamentally different models with enhanced social capabilities—able to grasp subtle nuances in human communication and adapt more fluidly to conversational contexts.

This development prompted me to link it to the nature of intelligence itself. Human intelligence isn't purely cognitive; it's embodied. We think with our bodies, feel with our minds, and understand through physical experience. Many of our expressions reflect this reality: we "touch base," "feel our way through problems," and "get in touch with" ideas. Intelligence emerges from the interplay between thinking and experiencing, between abstract processing and physical sensation.

This raises questions about artificial social intelligence. Can AI truly develop sophisticated social intelligence without some form of embodiment? To understand human emotions and social dynamics deeply, might AI need to experience what it means to exist physically in the world? To have a body that moves through space, interacts with environments, and feels physical responses?

Perhaps the next frontier in AI isn't just about creating systems that can handle tasks too dull or difficult for humans but about developing intelligences that can meaningfully connect with our embodied experience. A truly socially intelligent AI might need to understand not just what we say, but how it feels to be human—to move, touch, and physically respond to others.

The rapid and widespread diffusion of generative artificial intelligence (AI) has unlocked new capabilities and changed how content and services are created, shared, and consumed.

The article then highlights a comprehensive framework to govern generative AI, emphasizing the need for adaptive, participatory, and proactive approaches.

Araz Taeihagh, Governance of Generative AI, Policy and Society, 2025;, puaf001, https://doi.org/10.1093/polsoc/puaf001

What are the plans for the coming week?

This evening I hope I can join the State of Design 2025. The Mobile World Conference in Barcelona is happening again. Curious if some news will get traction here.

I will also for sure keep an eye on all posts on this years SXSW, especially these Dutchies. The last time I was there myself was back in 2019, I did not took the plane there after covid anymore, but with seven visits I feel still a bit connected from a distance :-)

And checking out relevant projects and people for the coming activities and events.

See you next week!

References with the notions

For the Monthly Cities of Things update I wrote a reflection inspired by last months news shared here. This time is packed with speculative things of the near future.

Human-AI partnerships

What will the impact of outsourcing thinking be? A cognitive wilderness.

With the introduction of ChatGPT 4.5 and Claude 3.7, the focus on reasoning is combined with models with more social intelligence. Grok-3 is doing this differently. AI bullying…

Gemini is introducing a new feature on live video

Alexa is finally catching up with AI. Or is it?

AGI needs some human investing first.

Bad vibes for Siri

These explorations into possible products by Matt are always very insightful as a reference for directions (in product development, not in wayfinding).