Post by swamprat on Nov 9, 2017 18:36:14 GMT -6
Four ethical priorities for neurotechnologies and AI
BY: Rafael Yuste, Sara Goering, Blaise Agüera y Arcas, Guoqiang Bi, Jose M. Carmena, Adrian Carter, Joseph J. Fins, Phoebe Friesen, Jack Gallant, Jane E. Huggins, Judy Illes, Philipp Kellmeyer, Eran Klein, Adam Marblestone, Christine Mitchell, Erik Parens, Michelle Pham, Alan Rubel, Norihiro Sadato, Laura Specker Sullivan, Mina Teicher, David Wasserman, Anna Wexler, Meredith Whittaker& Jonathan Wolpaw
08 November 2017
Artificial intelligence and brain–computer interfaces must respect and preserve people's privacy, identity, agency and equality, say Rafael Yuste, Sara Goering and colleagues.
A man with a spinal-cord injury (right) prepares for a virtual cycle race in which competitors steer avatars using brain signals.
Consider the following scenario. A paralysed man participates in a clinical trial of a brain–computer interface (BCI). A computer connected to a chip in his brain is trained to interpret the neural activity resulting from his mental rehearsals of an action. The computer generates commands that move a robotic arm. One day, the man feels frustrated with the experimental team. Later, his robotic hand crushes a cup after taking it from one of the research assistants, and hurts the assistant. Apologizing for what he says must have been a malfunction of the device, he wonders whether his frustration with the team played a part.
This scenario is hypothetical. But it illustrates some of the challenges that society might be heading towards.
Current BCI technology is mainly focused on therapeutic outcomes, such as helping people with spinal-cord injuries. It already enables users to perform relatively simple motor tasks — moving a computer cursor or controlling a motorized wheelchair, for example. Moreover, researchers can already interpret a person's neural activity from functional magnetic resonance imaging scans at a rudimentary level — that the individual is thinking of a person, say, rather than a car.
It might take years or even decades until BCI and other neurotechnologies are part of our daily lives. But technological developments mean that we are on a path to a world in which it will be possible to decode people's mental processes and directly manipulate the brain mechanisms underlying their intentions, emotions and decisions; where individuals could communicate with others simply by thinking; and where powerful computational systems linked directly to people's brains aid their interactions with the world such that their mental and physical abilities are greatly enhanced.
Such advances could revolutionize the treatment of many conditions, from brain injury and paralysis to epilepsy and schizophrenia, and transform human experience for the better. But the technology could also exacerbate social inequalities and offer corporations, hackers, governments or anyone else new ways to exploit and manipulate people. And it could profoundly alter some core human characteristics: private mental life, individual agency and an understanding of individuals as entities bound by their bodies.
It is crucial to consider the possible ramifications now.
The Morningside Group comprises neuroscientists, neurotechnologists, clinicians, ethicists and machine-intelligence engineers. It includes representatives from Google and Kernel (a neurotechnology start-up in Los Angeles, California); from international brain projects; and from academic and research institutions in the United States, Canada, Europe, Israel, China, Japan and Australia. We gathered at a workshop sponsored by the US National Science Foundation at Columbia University, New York, in May 2017 to discuss the ethics of neurotechnologies and machine intelligence.
We believe that existing ethics guidelines are insufficient for this realm. These include the Declaration of Helsinki, a statement of ethical principles first established in 1964 for medical research involving human subjects (go.nature.com/2z262ag); the Belmont Report, a 1979 statement crafted by the US National Commission for the Protection of Human Subjects of Biomedical and Behavioural Research (go.nature.com/2hrezmb); and the Asilomar artificial intelligence (AI) statement of cautionary principles, published early this year and signed by business leaders and AI researchers, among others (go.nature.com/2ihnqac).
To begin to address this deficit, here we lay out recommendations relating to four areas of concern: privacy and consent; agency and identity; augmentation; and bias. Different nations and people of varying religions, ethnicities and socio-economic backgrounds will have differing needs and outlooks. As such, governments must create their own deliberative bodies to mediate open debate involving representatives from all sectors of society, and to determine how to translate these guidelines into policy, including specific laws and regulations.
Intelligent investments
Some of the world's wealthiest investors are betting on the interplay between neuroscience and AI. More than a dozen companies worldwide, including Kernel and Elon Musk's start-up firm Neuralink, which launched this year, are investing in the creation of devices that can both 'read' human brain activity and 'write' neural information into the brain. We estimate that current spending on neurotechnology by for-profit industry is already US$100 million per year, and growing fast.
Investment from other sectors is also considerable. Since 2013, more than $500 million in federal funds has gone towards the development of neurotechnology under the US BRAIN initiative alone.
Current capabilities are already impressive. A neuroscientist paralysed by amyotrophic lateral sclerosis (ALS; also known as Lou Gehrig's or motor neuron disease) has used a BCI to run his laboratory, write grant applications and send e-mails. Meanwhile, researchers at Duke University in Durham, North Carolina, have shown that three monkeys with electrode implants can operate as a 'brain net' to move an avatar arm collaboratively. These devices can work across thousands of kilometres if the signal is transmitted wirelessly by the Internet.
Soon such coarse devices, which can stimulate and read the activity of a few dozen neurons at most, will be surpassed. Earlier this year, the US Defense Advanced Research Projects Agency (DARPA) launched a project called Neural Engineering System Design. It aims to win approval from the US Food and Drug Administration within 4 years for a wireless human brain device that can monitor brain activity using 1 million electrodes simultaneously and selectively stimulate up to 100,000 neurons.
Meanwhile, Google, IBM, Microsoft, Facebook, Apple and numerous start-ups are building ever-more-sophisticated artificial neural networks that can already outperform humans on tasks with well-defined inputs and outputs.
Last year, for example, researchers at the University of Washington in Seattle demonstrated that Google's FaceNet system could recognize one face from a million others. Another Google system with similar neural-network architecture far outperforms well-travelled humans at guessing where in the world a street scene has been photographed, demonstrating the generality of the technique. In August, Microsoft announced that, in certain metrics, its neural network for recognizing conversational speech has matched the abilities of even trained professionals, who have the option of repeatedly rewinding and listening to words used in context. And using electroencephalogram (EEG) data, researchers at the University of Freiburg in Germany showed in July how neural networks can be used to decode planning-related brain activity and so control robots.
Future neural networks derived from a better understanding of how real ones work will almost certainly be much more powerful even than these examples. The artificial networks in current use have been inspired by models of brain circuits that are more than 50 years old, which are based on recording the activity of individual neurons in anaesthetized animals. In today's neuroscience labs, researchers can monitor and manipulate the activity of thousands of neurons in awake, behaving animals, owing to advances in optical methods, computing, molecular engineering and microelectronics.
We are already intimately connected to our machines. Researchers at Google calculated this year that the average user touches their phone nearly one million times annually (unpublished data). The human brain controls auditory and visual systems to decipher sounds and images, and commands limbs to hold and manipulate our gadgets. Yet the convergence of developments in neurotechnologies and AI would offer something qualitatively different — the direct linking of people's brains to machine intelligence, and the bypassing of the normal sensorimotor functions of brains and bodies.
Read more: www.nature.com/news/four-ethical-priorities-for-neurotechnologies-and-ai-1.22960?WT.ec_id=NEWSDAILY-20171109&utm_source=briefing&utm_medium=email&utm_campaign=20171109
BY: Rafael Yuste, Sara Goering, Blaise Agüera y Arcas, Guoqiang Bi, Jose M. Carmena, Adrian Carter, Joseph J. Fins, Phoebe Friesen, Jack Gallant, Jane E. Huggins, Judy Illes, Philipp Kellmeyer, Eran Klein, Adam Marblestone, Christine Mitchell, Erik Parens, Michelle Pham, Alan Rubel, Norihiro Sadato, Laura Specker Sullivan, Mina Teicher, David Wasserman, Anna Wexler, Meredith Whittaker& Jonathan Wolpaw
08 November 2017
Artificial intelligence and brain–computer interfaces must respect and preserve people's privacy, identity, agency and equality, say Rafael Yuste, Sara Goering and colleagues.
A man with a spinal-cord injury (right) prepares for a virtual cycle race in which competitors steer avatars using brain signals.
Consider the following scenario. A paralysed man participates in a clinical trial of a brain–computer interface (BCI). A computer connected to a chip in his brain is trained to interpret the neural activity resulting from his mental rehearsals of an action. The computer generates commands that move a robotic arm. One day, the man feels frustrated with the experimental team. Later, his robotic hand crushes a cup after taking it from one of the research assistants, and hurts the assistant. Apologizing for what he says must have been a malfunction of the device, he wonders whether his frustration with the team played a part.
This scenario is hypothetical. But it illustrates some of the challenges that society might be heading towards.
Current BCI technology is mainly focused on therapeutic outcomes, such as helping people with spinal-cord injuries. It already enables users to perform relatively simple motor tasks — moving a computer cursor or controlling a motorized wheelchair, for example. Moreover, researchers can already interpret a person's neural activity from functional magnetic resonance imaging scans at a rudimentary level — that the individual is thinking of a person, say, rather than a car.
It might take years or even decades until BCI and other neurotechnologies are part of our daily lives. But technological developments mean that we are on a path to a world in which it will be possible to decode people's mental processes and directly manipulate the brain mechanisms underlying their intentions, emotions and decisions; where individuals could communicate with others simply by thinking; and where powerful computational systems linked directly to people's brains aid their interactions with the world such that their mental and physical abilities are greatly enhanced.
Such advances could revolutionize the treatment of many conditions, from brain injury and paralysis to epilepsy and schizophrenia, and transform human experience for the better. But the technology could also exacerbate social inequalities and offer corporations, hackers, governments or anyone else new ways to exploit and manipulate people. And it could profoundly alter some core human characteristics: private mental life, individual agency and an understanding of individuals as entities bound by their bodies.
It is crucial to consider the possible ramifications now.
The Morningside Group comprises neuroscientists, neurotechnologists, clinicians, ethicists and machine-intelligence engineers. It includes representatives from Google and Kernel (a neurotechnology start-up in Los Angeles, California); from international brain projects; and from academic and research institutions in the United States, Canada, Europe, Israel, China, Japan and Australia. We gathered at a workshop sponsored by the US National Science Foundation at Columbia University, New York, in May 2017 to discuss the ethics of neurotechnologies and machine intelligence.
We believe that existing ethics guidelines are insufficient for this realm. These include the Declaration of Helsinki, a statement of ethical principles first established in 1964 for medical research involving human subjects (go.nature.com/2z262ag); the Belmont Report, a 1979 statement crafted by the US National Commission for the Protection of Human Subjects of Biomedical and Behavioural Research (go.nature.com/2hrezmb); and the Asilomar artificial intelligence (AI) statement of cautionary principles, published early this year and signed by business leaders and AI researchers, among others (go.nature.com/2ihnqac).
To begin to address this deficit, here we lay out recommendations relating to four areas of concern: privacy and consent; agency and identity; augmentation; and bias. Different nations and people of varying religions, ethnicities and socio-economic backgrounds will have differing needs and outlooks. As such, governments must create their own deliberative bodies to mediate open debate involving representatives from all sectors of society, and to determine how to translate these guidelines into policy, including specific laws and regulations.
Intelligent investments
Some of the world's wealthiest investors are betting on the interplay between neuroscience and AI. More than a dozen companies worldwide, including Kernel and Elon Musk's start-up firm Neuralink, which launched this year, are investing in the creation of devices that can both 'read' human brain activity and 'write' neural information into the brain. We estimate that current spending on neurotechnology by for-profit industry is already US$100 million per year, and growing fast.
Investment from other sectors is also considerable. Since 2013, more than $500 million in federal funds has gone towards the development of neurotechnology under the US BRAIN initiative alone.
Current capabilities are already impressive. A neuroscientist paralysed by amyotrophic lateral sclerosis (ALS; also known as Lou Gehrig's or motor neuron disease) has used a BCI to run his laboratory, write grant applications and send e-mails. Meanwhile, researchers at Duke University in Durham, North Carolina, have shown that three monkeys with electrode implants can operate as a 'brain net' to move an avatar arm collaboratively. These devices can work across thousands of kilometres if the signal is transmitted wirelessly by the Internet.
Soon such coarse devices, which can stimulate and read the activity of a few dozen neurons at most, will be surpassed. Earlier this year, the US Defense Advanced Research Projects Agency (DARPA) launched a project called Neural Engineering System Design. It aims to win approval from the US Food and Drug Administration within 4 years for a wireless human brain device that can monitor brain activity using 1 million electrodes simultaneously and selectively stimulate up to 100,000 neurons.
Meanwhile, Google, IBM, Microsoft, Facebook, Apple and numerous start-ups are building ever-more-sophisticated artificial neural networks that can already outperform humans on tasks with well-defined inputs and outputs.
Last year, for example, researchers at the University of Washington in Seattle demonstrated that Google's FaceNet system could recognize one face from a million others. Another Google system with similar neural-network architecture far outperforms well-travelled humans at guessing where in the world a street scene has been photographed, demonstrating the generality of the technique. In August, Microsoft announced that, in certain metrics, its neural network for recognizing conversational speech has matched the abilities of even trained professionals, who have the option of repeatedly rewinding and listening to words used in context. And using electroencephalogram (EEG) data, researchers at the University of Freiburg in Germany showed in July how neural networks can be used to decode planning-related brain activity and so control robots.
Future neural networks derived from a better understanding of how real ones work will almost certainly be much more powerful even than these examples. The artificial networks in current use have been inspired by models of brain circuits that are more than 50 years old, which are based on recording the activity of individual neurons in anaesthetized animals. In today's neuroscience labs, researchers can monitor and manipulate the activity of thousands of neurons in awake, behaving animals, owing to advances in optical methods, computing, molecular engineering and microelectronics.
We are already intimately connected to our machines. Researchers at Google calculated this year that the average user touches their phone nearly one million times annually (unpublished data). The human brain controls auditory and visual systems to decipher sounds and images, and commands limbs to hold and manipulate our gadgets. Yet the convergence of developments in neurotechnologies and AI would offer something qualitatively different — the direct linking of people's brains to machine intelligence, and the bypassing of the normal sensorimotor functions of brains and bodies.
Read more: www.nature.com/news/four-ethical-priorities-for-neurotechnologies-and-ai-1.22960?WT.ec_id=NEWSDAILY-20171109&utm_source=briefing&utm_medium=email&utm_campaign=20171109