Robophilosophy: Philosophy of, for, and by Social Robotics is the first comprehensive book on robophilosophy, the field that studies what social robots do to how people relate, decide and see themselves. MIT Press published it in January 2026. The paperback costs 140 dollars, but the University of Helsinki offers the full published version as a free PDF under a Creative Commons licence. For any organisation planning to put a humanoid or social robot in front of customers, patients or staff, it is the most useful briefing you can get for free, because it names the social effects that European rules are starting to regulate.
This article explains what robophilosophy is, what is in the book, five concepts that translate into concrete questions for a deployment, and where the book falls short.
Why European organisations should read it now
Social robots are leaving the lab. 1X opened a factory in California for its home humanoid NEO and plans first customer deliveries in 2026. Robots that read intent, remember people and react to them are moving into care, retail and hospitality, as we saw with Ludi 0.2 and Europe’s rules for social robots and in our outlook on physical AI in the next decade.
European law now treats the social side of robots as a design matter. From 20 January 2027, the Machinery Regulation asks that machinery operating with varying levels of autonomy responds to people adequately and appropriately, for example verbally or through gestures, facial expressions or body movement, and that it communicates what it plans to do. The same ergonomics section asks manufacturers to reduce psychological stress from interacting with the machine. Since February 2025, the AI Act has banned AI systems that use manipulative or deceptive techniques to materially distort behaviour in ways that cause significant harm.
Both rules assume you can describe how people experience a robot and how the robot steers them. That vocabulary is exactly what this book offers. It is also a European product: the editors work at Aarhus University in Denmark and the University of Helsinki in Finland, and the field grew out of a conference series that started in Aarhus in 2014.
What robophilosophy is
Philosopher Johanna Seibt of Aarhus University coined the term in 2013. She describes robophilosophy as a systematic reconfiguration of philosophy in the face of artificial social agency. The biennial Robophilosophy conferences typically bring together 200 to 400 participants and 80 to 120 research contributions. The 2026 edition in Dublin ran under the theme “Connected Futures” and explicitly invited work on procedures for responsible technology development in relation to the EU AI Act.
The field works in three directions, which give the book its subtitle.

- Philosophy of social robotics is the reflective role. It asks what social robots mean for ethics, culture and society.
- Philosophy for social robotics puts philosophers at the design table, working with engineers and psychologists on what a robot should do and how to describe what happens between person and machine.
- Philosophy by social robotics runs the other way. Robots put pressure on concepts such as mind, consciousness, agency and autonomy, and force philosophy to revise them.
The editors’ preface to the 2024 conference proceedings describes how the field moved from the first to the second direction from 2016 onwards, with the third gaining attention since 2020 as AI capabilities grew.
What is in the book
The book is an edited volume of 498 pages by researchers from philosophy, psychology and the technical sciences. Johanna Seibt edited it with Raul Hakli of the University of Helsinki and Marco Nørskov of Aarhus University. After an introductory chapter by Seibt and Hakli, three parts follow the three directions. The table follows the table of contents in the publisher’s listing.
| Part | Core question | Chapters and authors |
|---|---|---|
| Philosophy of social robotics | What do social robots mean for morality and society? | The Machine Question (David Gunkel) · The Automation of the Social (Mark Coeckelbergh) · Who is Afraid of Nudging by Robots? (Raffaele Rodogno) · Robots, Empowerment and Equity (Illah Nourbakhsh) |
| Philosophy for social robotics | How do you help design robots that work socially? | Taking a Social Stance towards Robots (Raul Hakli) · OASIS (Johanna Seibt, Christina Vestergaard and Malene Damholdt) · Robot Sociality? (Mark Bickhard) · Dispersed Robots and Intelligent Environments (Jens Christian Bjerring and Jacob Busch) · Artificial Phronesis (John Sullins) |
| Philosophy by social robotics | What do robots teach us about humans? | Ethical Common Sense for Robots (Benjamin Kuipers) · Robotics as a new science of human beings (Domenico Parisi and Stefano Nolfi) |
Several chapters trace back to talks at the conference series. Mark Coeckelbergh spoke about the automation of the social at the first edition in 2014, under almost the same title. The volume collects more than a decade of work by one research community.
Five concepts, and the questions they raise for a deployment
The full chapters are in the PDF. Below are five core ideas, each explained through earlier published work by the same authors, followed by the question it puts on the table for an organisation that deploys a social robot.
1. Sociomorphing, not anthropomorphizing
The common explanation is that people humanise robots: they see a face and imagine a person behind it. Seibt, Vestergaard and Damholdt challenge that assumption. In their view, people often respond to actual social capacities of the robot that are not human, such as orienting towards you, taking turns or responding to what you do. They call this sociomorphing, and their OASIS framework, the Ontology of Asymmetric Social Interactions, describes these one-sided interactions. In the book it gets a chapter of almost sixty pages.

The question for your pilot: which social signals does the robot send, and do users read them as intended? That is the same ground the Machinery Regulation covers when it asks machines to respond appropriately and communicate their planned actions. Turn-taking is also hard to engineer, as two humanoids learning table tennis doubles showed.
2. Taking a social stance
Raul Hakli builds on Daniel Dennett’s intentional stance, the habit of predicting behaviour by attributing beliefs and desires. Hakli extends it into a social stance: people can coordinate with a robot by attributing certain social capacities to it for practical purposes, without believing it is a person.
The question for your pilot: which capacities do staff and customers need to attribute to the robot to work with it, and where does that attribution turn into misplaced trust? The answer belongs in the training material. Under Article 4 of the AI Act, deployers must take measures that support the AI literacy of the people who operate AI systems on their behalf, a duty the Digital Omnibus reworded in July 2026 but kept.
3. Nudging by robots
A robot that reminds a resident to drink water, or suggests a second coffee, steers behaviour. Raffaele Rodogno examines whether that is acceptable. In an earlier talk he argued that liberalism has no single answer: it depends on how autonomy is understood and how much weight it carries as a political value.
The question for your pilot: where does the robot steer choices, who benefits, and how would you show it stays well below the AI Act’s line on manipulation? Social robots are good at triggering reflexes such as gratitude, politeness and even tipping. That makes nudging a governance topic, not a marketing feature.
4. Artificial phronesis
Phronesis is Aristotle’s term for practical wisdom, the skill of doing the right thing in a situation you have never met before. John Sullins of Sonoma State University argues that designers of social robots have to solve this problem, or their project to bring robots into daily life will fail. The skill lets an agent quickly separate the many inappropriate actions from the few right ones.
The question for your pilot: which situations require judgement the robot does not have, and how does it hand those over to a person? Escalation design is where practical wisdom becomes an operational procedure.
5. Empowerment and equity
Illah Nourbakhsh, a roboticist at Carnegie Mellon University, contributes the chapter on robots, empowerment and equity. The title frames the question every deployment raises sooner or later: who gains power when a robot arrives, and who loses it? Think of care workers whose tasks change, older residents who receive a robot instead of a visit, or customers who cannot use the interface.
The question for your pilot: who is affected, who was consulted, and how will you measure whether the robot widened or narrowed the gap? The Robophilosophy 2026 conference listed the future of human work and human-robot co-working among its main topics.
| Concept from the book | Question for your deployment | Where EU rules already touch it |
|---|---|---|
| Sociomorphing | Which social signals does the robot send, and are they read as intended? | Machinery Regulation, Annex III 1.1.6, from 20 January 2027 |
| Social stance | What must people attribute to the robot to work with it, and where does trust go too far? | AI literacy measures under Article 4 of the AI Act |
| Nudging | Where does the robot steer choices, and for whose benefit? | AI Act ban on harmful manipulation, Article 5(1)(a) |
| Artificial phronesis | Which situations need human judgement, and how are they escalated? | Human oversight duties for high-risk AI systems, Article 14 of the AI Act |
| Empowerment and equity | Who gains and who loses, and who was consulted? | Information and consultation of workers under EU and national law |
For the timing of the AI Act obligations after the Digital Omnibus, see our EU AI Act timeline. For physical safety, the relevant standard is ISO 25785-1 for humanoid robots.
Where the book falls short
This is an academic volume, written for researchers, designers and policymakers. It is not a deployment manual, and the link to regulation and procurement is one you have to make yourself, as we did above.
Several chapters go back to talks given between 2014 and 2020, before the breakthrough of generative AI. The editors wrote in 2024 that multimodal generative AI suddenly brought the vision of ubiquitous social robots much closer. Robots that talk through a language model, or that a remote operator steers through unfamiliar tasks as NEO does, raise new questions. Read the book for its concepts. They age more slowly than the hardware.
Finally, the contributors in the table of contents are European and North American researchers. Organisations deploying robots in other cultural settings will want additional perspectives on what counts as appropriate social behaviour.
How to use it in your organisation
- Product and design teams should start with Part 2, especially the OASIS chapter and Hakli’s chapter on the social stance.
- Operations, care and customer service leads should read Rodogno on nudging and Nourbakhsh on empowerment and equity before signing a pilot.
- Compliance and ethics teams can use the five questions above as a first checklist next to the AI Act and the Machinery Regulation.
- Executives can read Chapter 1, in which Seibt and Hakli introduce the field. It is the fastest route to the vocabulary.
The free PDF is available through the University of Helsinki research portal, and the official edition through MIT Press via DOI 10.7551/mitpress/10777.001.0001. For a broader reflection on how old our questions about thinking machines are, see what Homer might have thought of AI.
Frequently asked questions
Is Robophilosophy free to read?
Yes. The University of Helsinki offers the full published version as a PDF under a CC BY-NC-ND licence. You may read and share it with attribution, for non-commercial purposes and without changes. The paperback costs 140 dollars.
Who edited the book?
Johanna Seibt and Marco Nørskov of Aarhus University and Raul Hakli of the University of Helsinki. Contributors include David Gunkel, Mark Coeckelbergh, John Sullins and Illah Nourbakhsh.
What is sociomorphing?
Sociomorphing means perceiving real, non-human social capacities in a robot, such as attention or turn-taking. Anthropomorphizing means projecting human qualities that are not there.
Is the book relevant for companies, or only for academics?
It is written for academics, but its concepts map directly onto questions companies face when they deploy social robots, from user training to manipulation risks under the AI Act.
Your customers will tip the robot
People thank robots, apologise to them and sometimes tip them. It is tempting to treat that as a quirk of early adopters. Robophilosophy suggests the opposite: these reflexes respond to real signals the machine sends, and those signals are designed. The organisations that do well with social robots will be the ones that decide on purpose which reflexes their robot may trigger, and which it should leave alone.





