A.I. Law: Ethical and Legal Dimensions of Artificial Intelligence

Paper ID:

485

Last updated: 10/05/2011 13:57:36

Criteria:

Impact:

Likelihood:

Controversy:

Where: Regional

When: 11-20yrs

How Fast: Unknown

0 people thought this paper expanded their thinking bullet

Keywords:

robotics, ethics, rights, law, artificial intelligence

Summary

Technological advances together with a growing understanding of the workings of the human brain could enable computer-based intelligence to exceed human capabilities in many areas within a few decades. With such technological advances in the field of artificial intelligence (A.I.) gathering pace, scientists, lawyers and ethicists alike are beginning to consider the societal implications of the creation of self-aware A.I.

Discussion

Currently A.I. is confined to robots and machines which are passive inanimate objects without rights or duties. If and when A.I. technology is developed that is ‘self-aware’, calls may be made for the extension to robots of various rights and responsibilities that have traditionally been limited to organic life-forms such as humans and animals.

If artificial intelligence is developed to a level where it can be deployed widely - some argue this is likely in the coming years - the debate may intensify. [1] Under present law, A.I. technology is just inanimate, passive property without rights or duties. [1] Computers and other advanced machines are not legal persons and have no legal standing. In other words, they cannot be sued or prosecuted but at the same time they cannot sue or launch complaints against others. Their owners, however, can act to protect their property rights.

Today, over one million household robots, and a further 1.1 million industrial robots, are operating worldwide. [2] These robots are largely used to perform tasks that require great levels of precision, or are hazardous, repetitive and/or boring. Some have no functional value at all such as robotic pets. Current advances in A.I. have seen robots being used in dangerous situations such as battlefields, and exploration of hazardous places such as the deep sea floor, volcanoes and space. A.I. is now also routinely used in complex surgical procedures such as brain and heart bypass surgery. [2] As the sophistication of A.I. grows, machines may develop with more complex, human-like characteristics and abilities. In such circumstances they may come to be viewed as active ‘agents’ or possibly ‘quasi-agents’ in the eyes of the law. [3] As agents under law they will also inherit certain rights and responsibilities.

The legal and ethical dimensions of A.I. are strongly linked. Scientists such as Peter Asaro at the Institute of Electrical and Electronics Engineers (IEEE) have recently been theorising about this link. They argue that ethical considerations, such as making sure A.I. is programmed to act in an ethical way, making sure the ethics of people who design and use A.I. technology are sound, and ensuring people treat A.I. agents in an ethical manner are increasingly being seen in terms of legal responsibilities. If self-aware A.I. agents do indeed become more ubiquitous in the future, legal theory provides the framework for considering responsibility and agency: for example, considering who should bear responsibility in the event of A.I. harming humans physically, mentally or financially.

Key considerations in the application of legal theory and moral philosophy to A.I. are the problem of defining what constitutes a ‘self-aware’ A.I. agent and what laws and rights should apply. Perhaps the pioneer of such schools of thought was the science-fiction author Isaac Asimov. In a 1942 short story ‘Runaround’ Asimov introduced three laws of robotics: a robot may not injure a human being or, through inaction, allow a human being to come to harm; a robot must obey orders given to it by human beings except where such orders would conflict with the First Law; and
a robot must protect its own existence as long as such protection does not conflict with the First or Second Law. With advances in A.I. gathering pace, such ideas are moving from the realms of science fiction into serious scientific and ethical debate.

Questions are being asked such as, should A.I. agents or quasi-agents be given partial (or diminished) responsibility for their actions? Consequently, should they only have partial rights as well? Another consideration is how responsibility is transferred between humans and A.I. agents. For example, considering how to prevent humans from unjustly attributing responsibility for their actions onto A.I. agents or deciding whether to charge an A.I. agent’s programmer or owner with negligence if an A.I. agent causes damage or breaks the law.

Perhaps the most interesting current issue where such legal debates are coming to the fore is the US Department of Defence’s increasing reliance on robotic weaponry. [4] Unmanned Aerial Vehicles (UAVs) and ground-based Sword robots are already routinely used in military operations in the Middle East. These are, at present, controlled by distant human operators; however, the Pentagon wants to give them more autonomy. This autonomy critically includes the decision on when to use lethal force, which will raise far-reaching ethical and legal questions. For example, current international and humanitarian legislation and established uniform codes of military conduct such as the Geneva Conventions, do not account for independently acting A.I. agents in warfare. [3]

The Pentagon has said that such machines programmed to make morally sound decisions would make far fewer mistakes than humans in wartime situations as they are unaffected by the stress of their surrounding environment. However, critics feel that such machines will inevitably make mistakes and, furthermore, what should be defined as ethically acceptable situations for the use of lethal force, and indeed who makes this decision?

Also on the horizon is automated trading in financial markets involving negotiation between artificially intelligent software trading agents. [5] Over the next 20 years more and more financial decisions will be delegated to A.I. software agents. A legal framework for division of profits gained by A.I. and even responsibility for losses incurred through A.I. trading will need to be considered.

Human interest and advocacy on behalf of animals has continued to rise in recent years. More and more aspects of human existence are becoming reliant on A.I., computers and machines, which has raised the question of whether A.I. agents in turn should be given rights. For instance, when A.I. technology manages to create a machine that shows comparable or even superior cognitive functioning to that of animals or even humans, should we extend to these machines the same rights? [6] Conversely, in the far future, should artificial intelligence ever surpass the intelligence of humans, power could shift in favour of the A.I. technology that we create? Super-intelligent machines may even have a greater capacity for all cognitive functions (including moral and ethical thought), [7] thus there may be a critical need to program them with human-friendly motivations. [8]

In fact, scientists in South Korea in August 2007 drew up a ‘Robot Ethics Charter’ which they hope will be approved by the South Korean Government. The key considerations of the charter are: ensuring that humans have control over robots, preventing illegal use of robots, protecting data acquired by robots and ensuring robots can be clearly identified and traced. [9] These South Korean scientists feel that a society where humans and robots live together will become a reality within the next 10 years, and such a charter is essential in order to ensure a harmonious coexistence between humans and robots. [9]

If this is the case, the emotional and behavioural dimension of this issue deserves consideration. We are already “involved” emotionally with our technology, as the proportion of those in surveys who claim they cannot live without their mobile phone suggests. It is likely that with the growing sophistication of A.I. will come increasing anthropomorphisation and emotional projection onto machine based intelligences.

Many examples of this anthropomorphisation come from the current utilisation of A.I. by the US Army. In one instance, during a live fire test at the Yuma Test Grounds in Arizona, a test involving a 5-foot long autonomous mine-clearing robot was called to a halt by an observing army colonel, who described the repeated exposure of the robot to damaging explosions as “inhumane” [21]. Many more examples come from the wars in Afghanistan and Iraq where robots charged with carrying out dangerous tasks such as reconnaissance and bomb disposal are given honorary ‘battlefield promotions’ and ‘distinguished service’ medals by the soldiers who depend on them. [21]

Another early indication of this shift can be found in the work of the Centre for Social Robotics at the Australian Centre for Field Robotics, an exhibition space dedicated to understanding and realising human-robot interaction in social environments. A recent example of their work is the ‘Fish-Bird’ project, an interactive auto-kinetic artwork consisting of two robots, in the form of wheelchairs, programmed to ‘communicate’ with each other and with their audience through movement and written text. When interacting with these robots, observers are encouraged to attribute human emotions and motivations to what they observe the robots doing such as ‘falling in love’, or ‘becoming more comfortable with observers.’ [10]

Implications

A monumental shift could occur if A.I. agents continue to be developed to the point where they can reproduce, or improve themselves. Self-improvement of A.I. agents could be evolutionary in nature. Once built, an A.I. agent's program could potentially continue to learn and adapt through its own experiments with the world.

A.I. theorists such as Robert Freitas and Benjamin Sokis argue that advances in A.I. technology could open up complicated issues in a range of areas. If A.I. agents were granted rights, it could be argued that citizens' responsibilities might be extended to them [1] , [11] . While unlikely within a 50-year timescale, the extension of legal rights to A.I. agents could be followed by demands that A.I. agents be subject to certain responsibilities within society such as voting, payment of taxes, and military service [1] .

A.I. agents' rights would invariably clash with the property rights of their owners, and the extension of rights to robotic beings could be manipulated through programming and mechanical abilities at faster rates of reproduction than humans. Sokis (2005), with the support of Pollack (2005) say this could prompt spirited debates as to how humans, animals and A.I. agents rank with respect to rights and responsibilities in our world [11] ,[12] . This would include debates and qualifications on consciousness, intelligence and the capacity for emotional response.

The use of A.I agents may enable increased labour output and greater analytical capabilities that lead to greater human prosperity and an improvement of the human condition [1] . Sokis (2005) suggests that in the distant future, short of full and immediate emancipation of A.I. agents, legal authorities acting on behalf of A.I. agents could appeal to courts and make petitions that they be granted ever-increasing rights until granted full legal rights under the law [11] . This is unlikely to occur in all countries simultaneously, meaning that the owners of A.I. agents may move them to jurisdictions more friendly to their property rights and less sympathetic to the rights of A.I. agents.

Such developments could advance arguments which hold that the human brain is nothing more than a special type of computer. This may challenge religious belief systems which hold that humans have a special place in creation as possessing mysterious, spiritual and divine properties. The extension of rights to A.I. agents to reproduction, property and resources could create more strain on the environment in the form of energy, resource and space use and production of waste, depending on the nature of their requirements.

David Calverley, a lawyer from Arizona State University, has been a key contributor to the debate around the potential for A.I. rights. He suggests that in the distant future, humanlike A.I. agents that demonstrate both intelligence and emotions may eventually become indistinguishable from humans. Consequently, people may isolate themselves from contact with humans in preference for A.I. agents. A further issue is the potential devaluation of human life – if A.I. agents could demonstrate all human virtues, but without our imperfections, then why would we need humans? [6] Calverley adds that, on the other side of the coin, however, is the potential for robotics to improve our lives, rather than devalue them. A.I. agents taking over the more mundane functions of our lives will bring freedom and empowerment by enabling people to focus on the things that they do best, or what they enjoy most. [6]

Ray Kurzweil writing in The Futurist states that the most important future progression of A.I. is the application of nanotechnology to human cognitive functioning. [13] Tucker argues that by about 2030 nanobots (tiny microscopic machines) will begin to expand the capacity of our minds through speeding up inter-neuronal connections with high-speed virtual connections. This will allow us to greatly improve our pattern-recognition abilities, memories, and overall thinking capacity, as well as to directly interface with A.I. The technology will also provide wireless communication from one brain to another, a form of telepathy.

Early indicators

Advances in artificial intelligence, such as development of self awareness in A.I. [5], [14] and computer modelling the brain; commonsense reasoning in A.I.; [15]
Successful development of techniques so that humans and computers could be connected through neural implants resulting in technology becoming increasingly indistinguishable from the human; [1]
Creation of an ethical decision program; [7]
Robotic Bill of Rights and increased membership of ASPCR; [16]
A computer passing the Turing test; [17] Deployment of a fully autonomous robot in warfare. [4]
A.I. entity passes an A-level; A.I. as a major source of redundancy; A.I. entity given basic rights. [10]

Drivers & Inhibitors

Drivers: development of artificial intelligence; military/security requirements promote more investment in A.I. development.

Inhibitors: regulation of artificial intelligence technology.
Theoretical and technological limits to the development of A.I. [1]
Ethical objections; legal objections;
Resistance to A.I. making certain jobs redundant in the labour market. [8]

Parallels & Precedents

Animal rights movement;
Emancipation of slaves.
Legal punishments aimed at nonhuman agents, namely corporations.
Advances in nanotechnology, biotechnology and carbon computing help accelerate advances in A.I.
[18][1][3][4][19][6][7][11][20][5][14][15][17][16][8][10][2][12] [9][13]

Social Bookmarking

Comments

Date/Time

User

People who viewed this paper also viewed

ID	Headline
500	City Limits? The Future of London
406	The Taboo Solution: Can Population Management Be a Solution to Climate Change?
300	Understanding Complexity: How to Answer the Big Questions
432	Better Understanding of Quantum Theory
73	Leaving the “Ever-Closer Union”: Could Britain Withdraw From the EU?

Sources

Ref.	Publisher	Date	Title	Category
1	Other		Freitas, Robert A. (Jr) (1985) 'The Legal Rights of Robots', Student Lawyer 13 (January) :54-56	Tech
2	Other		[18] Pollack, Jordan (2005) Ethics for the Robot Age." Wired. Issue 13.01"	Tech
3	Other		Asaro, P., M. (2007) ‘Robots and Responsibility from a Legal Perspective’; Paper submitted to a Workshop on Roboethics, within the 2007 IEEE International Conference on Robotics and Automation (ICRA'07), in Rome, 10 to 14 April 2007	Tech
4	The Economist		The Economist (2007); June 7th; Robot wars - Military technology: Unmanned vehicles and robot soldiers are on the march. Can such machines be programmed to act ethically?	Tech
5	New Scientist		Biever, C. (2007) ‘Sense of self keeps robots on the move’. New Scientist. November; Issue 2579.	Tech
6	Other		Calverley, D., J., (2006) ‘Android science and animal rights, does an analogy exist?’ Connection Science. Vol 18. No 4. p. 403-417.	Tech
7	Other		Bostrom, N. (2003) Ethical Issues in Artificial Intelligence. Cognitive, Emotive and Ethical Aspects of Decision Making in Humans and in Artificial Intelligence, Vol 2(1) pp12-17	Tech
8	Other		The Centre for Social Robotics; Sydney, 2006;	Tech
9	Other		Kurzweil, R. (2006) Reinventing Humanity: The Future of Human-Machine Intelligence; The Futurist March-April 2006; pp39-46	Tech
10	New Scientist		Graham-Rowe, D. (2006) Instant Expert: Robots. Special report on the robotics revolution. New Scientist. 4th September 2006	Tech
11	Other		Soskis, Benjamin. (2005) 'Man and the Machines: It's Time to Start thinking about how we might grant legal rights to computers', Legal Affairs, January / February	Tech
12	Other		South Korea draws up code of ethics for robots; Psyorg - August 7th 2007.]	Tech
13	Other		Garreau, J (2007) Bots on The Ground - In the Field of Battle (Or Even Above It), Robots Are a Soldier's Best Friend. Washington Post, Sunday, May 6, 2007; D01.	Tech
14	The Economist		I think, therefore I am, I think - Consciousness awaits its Einstein. The Economist (2006); December 19th	Tech
15	Other		Petit, C.; Haseltine, E.C.; Engelberger, J.; Kawamura, K.; Thrun, S; Minsky, M.; McCarthy, J.; Herkert, J.; & Schultz, A.C. (2000) The Future of Humanoid Robots	Tech
16	Other		Sotala, K. (2007) ‘Artificial intelligence within our lifetime? No idle speculation’ Posted on www.acceleratingfuture.com; 27th of July, 2007	Tech
17	The Economist		Arresting developments - Computer science and biological science have a lot to teach each other. The Economist (2007); July 12th	Tech
18	Other		Pollack, Jordan (2005) Ethics for the Robot Age." Wired. Issue 13.01."	Tech
19	Other		Trading agents-the new world of artificial intelligence. The Dolphin (November 2001)"	Tech
20	New Scientist		[9] Biever, C. (2007) ‘Self-aware robot turns mirror on humankind’. New Scientist. November; Issue 2604.	Tech

Other Sources

Link	User
[14] The American Society for Prevention of Cruelty to Robots	JT
[15] The Singularity Institute for Artificial Intelligence	JT
[16] BT Technology Timeline 2006-2051;	JT

The contents of this paper were provided by the Outsights-Ipsos MORI Partnership. Any views expressed are independent of government and do not constitute government policy.