BY DR ELEANOR DRAGE
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Dr Eleanor Drage is a Senior Research Fellow at the University of Cambridge's Leverhulme Centre for the Future of Intelligence. One of the UK's leading AI ethicists, she is the author of The Planetary Humanism of European Women’s SF: An Experience of the Impossible (2023), a history of science fiction written by women from 1666-2016. She's co-editor of Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines and the forthcoming book The Good Robot: Why Technologies of the Future Need Feminism with Bloomsbury. Eleanor is co-host of The Good Robot Podcast, where she interviews top scholars and technologists about AI ethics.
BY DR ELEANOR DRAGE
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Dr Eleanor Drage is a Senior Research Fellow at the University of Cambridge's Leverhulme Centre for the Future of Intelligence. One of the UK's leading AI ethicists, she is the author of The Planetary Humanism of European Women’s SF: An Experience of the Impossible (2023), a history of science fiction written by women from 1666-2016. She's co-editor of Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines and the forthcoming book The Good Robot: Why Technologies of the Future Need Feminism with Bloomsbury. Eleanor is co-host of The Good Robot Podcast, where she interviews top scholars and technologists about AI ethics.
BY DR ELEANOR DRAGE
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Dr Eleanor Drage is a Senior Research Fellow at the University of Cambridge's Leverhulme Centre for the Future of Intelligence. One of the UK's leading AI ethicists, she is the author of The Planetary Humanism of European Women’s SF: An Experience of the Impossible (2023), a history of science fiction written by women from 1666-2016. She's co-editor of Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines and the forthcoming book The Good Robot: Why Technologies of the Future Need Feminism with Bloomsbury. Eleanor is co-host of The Good Robot Podcast, where she interviews top scholars and technologists about AI ethics.
BY DR ELEANOR DRAGE
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Beyond dead and alive
When I first encountered the work of scholars who investigate why we should think of inanimate matter as ‘alive’, or at least ‘lively’, I was captivated. Jane Bennett’s descriptions of the trickle of chemical waste in junkyards are not unlike the figures that come to life in Toy Story (1995), where viewers also get a glimpse into the ‘secret life of things’.[1] She shows how there’s an aliveness in the sludge of chemical waste and the meeting of decaying materials by drawing attention to the interactions between chemicals that create ‘animacy’. The ‘inanimate’ world is, frankly, much more lively than we give it credit for. Take the catastrophic 2003 Northeast blackout in the US, which Bennett uses as an example of how interactions between parts of the energy infrastructure we might not see as ‘alive’ can be the agents of disaster.[2] The blackout was caused by a series of faults, from a power station suddenly going offline to a software bug that prevented operators from taking action to manage the system. This led to an overloaded power line coming into contact with trees and triggering a series of circuit breaks that cascaded into the worst power outage in American history. The blackout made it all too clear that non-human things have power in energy systems, and should therefore be taken seriously as actors in the network. This is also the case with human consciousness, a highly energy-intensive process which depends on a network of actors, not all of them animate or ‘human’. The doctrine of ‘human exceptionalism’, the idea that we are different/better than everything else, would have it that consciousness is a human-only affair. But as scholars like N. Katherine Hayles have demonstrated, there are lots of non-conscious processes that structure human cognition,[3] from our interactions with technology to the relationship between thought and cell function. Willed action has always been influenced by technologies – the things we use to cook, write or travel with – and phenomena like cell function that we may not be aware of but which make awareness possible. We therefore need to be ‘biophilic’,[4] Hayles says, in our approach to understanding what consciousness is, which means keeping our ear to the ground and being attuned to how humans are involved in other life forms. If we do this, we can explore consciousness in a way that enhances connections to the natural environment rather than wipes them away. So much of what we call consciousness is supported by non-human cognition, which means that there are far more interesting ways of thinking about the non-human world than as a set of objects and ‘objective’ machines. The wonderfully complex, sticky evolutionary history that humanity shares with non-conscious entities that nevertheless do ‘think’, from technology to bacteria, is often overlooked when people ask the question, ‘Can machines be conscious?’ People tend to imagine ‘machine consciousness’ as machines thinking and acting independently of humans, just as discussions around human consciousness overlook how decision-making processes are distributed across our bodies.
Mainstream biology is now cottoning on to the relationship between conscious and unconscious cognition that up until the early 2020s thrived at the margins of science. An entirely new discipline of nutritional psychiatry has emerged in response to evidence that the gut affects your mood, influencing how you handle stress and whether you experience mental ill-health. I like to see this as a surprising resurrection of elements of early medieval medical texts on the humours, wherein bodily fluids directly conjure states of mind, meaning that your irritability could be conveniently blamed on a yellowing of the bile. Much of what is considered ‘frontier’ science on the gut and mood was also preceded by twenty-first-century feminist philosophy of science, such as the work of Elizabeth Wilson, who over a decade ago was using feminist ideas to challenge mainstream views of what is and isn’t a ‘thinking’ (cognitive) process.[5] American ecologist and philosopher David Abram has also explored the moods of our senses, for example by describing smell and touch as ‘gregarious’ organs that participate in – and are dependent upon – the world around us.[6] I love this description, which takes seriously how distributed consciousness is. We don’t just process the world up there in our brains,but everywhere throughout our bodies. If physical processes – whether gastronomic or sensorial – involve and alter cognition, does that make the human brain less valuable? Equally,are thinking machines less awe-inspiring if the cognition takes place in many different parts of their distributed ‘bodies’ – in the data centres, in the humans doing the data-labelling, or in the quartz mined and manufactured for semiconductors?
When AI opinion leaders harp on about AI consciousness, they are often fetishising the brain (or computer as brain) as the location of intelligence. The obsession with the brain, or some central locus of intelligence, is pathetic and outdated,harking back to a time before we understood just how distributed cognition is, or appreciated that other animals did any thinking. There is no such thing as a neutral approach to cognition – how we understand what it means to think always says something about who we believe we are as humans in relation to other Earth-dwelling creatures. When we worry whether AI can be conscious we also show a concern for our place in the hierarchy of things. Theories of consciousness are as much full of philosophical, religious and humanistic ideas and ideals as they are scientific facts. I’m therefore on board to ditch the phrase ‘machine consciousness’, which makes machine cognition seem like a mystical artefact rather than a distributed material network, and the tiresome, pointless discussions that come with the assumption that all kinds of consciousness are the same. I am invested in ideas about consciousness – whether in relation to humans or machines – that are ecologically suggestive, socially beneficial and grounded in new data about the relationships between the bits of your body traditionally associated with thinking, like the brain, and those that are not, like the gut.
Some space in the dustbin should also be saved for another widely disputed term in AI, the idea of ‘machine autonomy’, often taken to mean machine ‘independence’. Disability theory shows that the experience of autonomy for a human depends on how accessible you find social systems and facilities. Similarly, when we hide the infrastructure that supports machine ‘autonomy’, we elevate AI from a product that needs engineering, monitoring and training to a god in machine form. It is never just software or applied statistics. When we interact with any AI technology we are participating in a system of engineers and data labellers,corporate infrastructure, crowdsourced workers, your own data and – often – that of all a technology’s users worldwide. The mouthpiece of this assemblage is the chatbot: capitalist magic turns behind-the-scenes labour into an autonomous product.
That’s also why AI isn’t ‘just a tool’: tools are assemblages,as we have seen, of human and non-human actors. From medicines to machines to multinational conglomerates, they construct worlds and change our minds and bodies, our societies and the environment. You would think that the more we learn about how technology shapes human cognition, the more we would avoid creating tools that are detrimental to the human psyche. Amid a disaster of contradictory messages about what it means to be human, conscious and autonomous in the age of technology, we need more risky and expansive responses to the question of what AI can be, and how we should be interacting with it. We must encourage perspectives that respect our inti-mate ties to the environment and recognise that to be human is also to be technological. Only then can we establish some continuity between who we were before AI and who we are now, and come up with some creative and less defeatist answers to the question of what will become of humanity and its symbionts in the future.
1. Jane Bennett, ‘The agency of assemblages and the North
American blackout’, Public Culture, 17:3 (2005), pp. 445–65.
2. N. Katherine Hayles, Unthought (Chicago: University of
Chicago Press, 2017).
3. N. Katherine Hayles, ‘Good Technology Is Biophilic’, in
Eleanor Drage & Kerry McInerney (eds), The Good Robot:
Why Technology Needs Feminism (London: Bloomsbury
Academic, 2024), pp. 28–35.
4. Elizabeth A. Wilson, Neural Geographies (New York:
Routledge, 1998), p. 6.
5. David Abram, ‘The ecology of perception – an interview’,
Emergence Magazine, 2018.
6. Tomasz Hollanek & Katarzyna Nowaczyk‐Basińska,
‘Griefbots, deadbots, postmortem avatars’, Philosophy &
Technology, 37 (2024), pp. 1–22.
This article is an extract from What If We Got AI Right?: How to stop catastrophising and build an ethical future by Eleanor Drage. Published 2026 by Profile Books.
Dr Eleanor Drage is a Senior Research Fellow at the University of Cambridge's Leverhulme Centre for the Future of Intelligence. One of the UK's leading AI ethicists, she is the author of The Planetary Humanism of European Women’s SF: An Experience of the Impossible (2023), a history of science fiction written by women from 1666-2016. She's co-editor of Feminist AI: Critical Perspectives on Algorithms, Data, and Intelligent Machines and the forthcoming book The Good Robot: Why Technologies of the Future Need Feminism with Bloomsbury. Eleanor is co-host of The Good Robot Podcast, where she interviews top scholars and technologists about AI ethics.