The initial AI-Ethics International Conference on Principles will include discussion of the principles already articulated by other groups. Of course we start with Issac Aismov and his famous 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 the orders given it by human beings except where such orders would conflict with the First Law.
- A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
But there is a lot more needed than that and some even contend Aismov had it wrong.
So far we have identified proposals from six different groups or individuals. Also See: European Civil Law Rules in Robotics (European Parliament’s Legal Affairs Committee); Position on Robotics and Artificial Intelligence (Proposal of Green Digital Working Group in European Parliament).
1. Future of Life Institute
One set of new proposed principles that we are impressed with are the twenty-three principles developed at the Future of Life 2017 Asilomar conference in Asilomar, California, called the AI Asilomar Principles. The Asilomar Conference was put on by the Future of Life Institute, which is based in Boston.
It was attended by an impressive group of: “AI researchers from academia and industry, and thought leaders in economics, law, ethics, and philosophy.” The conference covered a broad range of topics concerning beneficial AI, not just ethics. Here are the twenty-three Asilomar Principles:
1) Research Goal: The goal of AI research should be to create not undirected intelligence, but beneficial intelligence.
2) Research Funding: Investments in AI should be accompanied by funding for research on ensuring its beneficial use, including thorny questions in computer science, economics, law, ethics, and social studies, such as:
- How can we make future AI systems highly robust, so that they do what we want without malfunctioning or getting hacked?
- How can we grow our prosperity through automation while maintaining people’s resources and purpose?
- How can we update our legal systems to be more fair and efficient, to keep pace with AI, and to manage the risks associated with AI?
- What set of values should AI be aligned with, and what legal and ethical status should it have?
3) Science-Policy Link: There should be constructive and healthy exchange between AI researchers and policy-makers.
4) Research Culture: A culture of cooperation, trust, and transparency should be fostered among researchers and developers of AI.
5) Race Avoidance: Teams developing AI systems should actively cooperate to avoid corner-cutting on safety standards.
Ethics and Values
6) Safety: AI systems should be safe and secure throughout their operational lifetime, and verifiably so where applicable and feasible.
7) Failure Transparency: If an AI system causes harm, it should be possible to ascertain why.
8) Judicial Transparency: Any involvement by an autonomous system in judicial decision-making should provide a satisfactory explanation auditable by a competent human authority.
9) Responsibility: Designers and builders of advanced AI systems are stakeholders in the moral implications of their use, misuse, and actions, with a responsibility and opportunity to shape those implications.
10) Value Alignment: Highly autonomous AI systems should be designed so that their goals and behaviors can be assured to align with human values throughout their operation.
11) Human Values: AI systems should be designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity.
12) Personal Privacy: People should have the right to access, manage and control the data they generate, given AI systems’ power to analyze and utilize that data.
13) Liberty and Privacy: The application of AI to personal data must not unreasonably curtail people’s real or perceived liberty.
14) Shared Benefit: AI technologies should benefit and empower as many people as possible.
15) Shared Prosperity: The economic prosperity created by AI should be shared broadly, to benefit all of humanity.
16) Human Control: Humans should choose how and whether to delegate decisions to AI systems, to accomplish human-chosen objectives.
17) Non-subversion: The power conferred by control of highly advanced AI systems should respect and improve, rather than subvert, the social and civic processes on which the health of society depends.
18) AI Arms Race: An arms race in lethal autonomous weapons should be avoided.
19) Capability Caution: There being no consensus, we should avoid strong assumptions regarding upper limits on future AI capabilities.
20) Importance: Advanced AI could represent a profound change in the history of life on Earth, and should be planned for and managed with commensurate care and resources.
21) Risks: Risks posed by AI systems, especially catastrophic or existential risks, must be subject to planning and mitigation efforts commensurate with their expected impact.
22) Recursive Self-Improvement: AI systems designed to recursively self-improve or self-replicate in a manner that could lead to rapidly increasing quality or quantity must be subject to strict safety and control measures.
23) Common Good: Superintelligence should only be developed in the service of widely shared ethical ideals, and for the benefit of all humanity rather than one state or organization.
The Future of Life group has a website page for Discussion About the Asilomar AI Principles. We recommend that you check that out.
2. Institute of Electrical and Electronics Engineers
The Institute of Electrical and Electronics Engineers (IEEE) calls itself the world’s largest technical professional organization dedicated to advancing technology for the benefit of humanity. The IEEE Standards Association Global Initiative for Ethical Considerations in Artificial Intelligence and Autonomous Systems is active in the area of AI ethics. So far they have agreed upon four general principles.
- Human Benefit
- Education and Awareness
They have a very impressive, large group and we look forward to more work-product from the engineers soon.
3. Conference Toward AI Network Society
The proposed principles for AI Research and Development by the Conference Toward AI Network Society are also interesting. This group is led by Susumu Hirano, Professor of Law at Chuo University in Tokyo, Japan. Hirano has an LLM from Cornell Law School and is, or was, a member of the New York State Bar Association. See his presentation to the Carnegie Endowment for Internal Peace, Policy Issues Toward AI Networking and Guiding Principles for AI Development. They have proposed eight principles as a starting point:
Proposed Principles in “AI R&D Guideline”1. Principle of Transparency. Ensuring the abilities to explain and verify the behaviors of the AI network system.2. Principle of User Assistance. Giving consideration so that the AI network system can assist users and appropriately provide users with opportunities to make choices.3. Principle of Control-ability. Ensuring control-ability of the AI network system by humans.4. Principle of Security. Ensuring the robustness and dependability of the AI network system.5. Principle of Safety. Giving consideration so that the AI network system will not cause danger to the lives/bodies of users and third parties.6. Principle of Privacy. Giving consideration so that the AI network system will not infringe the privacy of users and third parties.7. Principle of Ethics. Respecting human dignity and individuals’ autonomy in conducting research and development of AI to be networked.8. Principle of Accountability. Accomplishing accountability to related stakeholders such as users by researchers/developers of AI to be networked.
We will be keeping our eye out for more work by Hirano and other groups in Japan. See: The Japanese Society for Artificial Intelligence Ethical Guidelines; Japan’s New Robot Strategy (10/2/15); Japan’s Report on Artificial Intelligence and Human Society (3/24/17)
4. Association of Computing Machinery
We also find interesting the January 12, 2017 Statement on Algorithmic Transparency and Accountability of the U.S. Public Policy Council of the Association of Computing Machinery (ACM) (USACM). It creates principles limited to the issue of AI bias. In 2017 the US Public Policy Council of the prestigious ACM agreed upon a set of seven principles to require Algorithmic Transparency and Accountability as part of AI regulation. In the Statement they explained the background:
The use of algorithms for automated decision making about individuals can result in harmful discrimination. Policymakers should hold institutions using analytics to the same standards as institutions where humans have traditionally made decisions and developers should plan and architect analytical systems to adhere to those standards when algorithms are used to make automated decisions or as input to decisions made by people.
Here are the seven principles the USACM formulated to help prevent AI bias:
Principles for Algorithmic Transparency and Accountability1. Awareness: Owners, designers, builders, users, and other stakeholders of analytic systems should be aware of the possible biases involved in their design, implementation, and use and the potential harm that biases can cause to individuals and society.2. Access and redress: Regulators should encourage the adoption of mechanisms that enable questioning and redress for individuals and groups that are adversely affected by algorithmically informed decisions.3. Accountability: Institutions should be held responsible for decisions made by the algorithms that they use, even if it is not feasible to explain in detail how the algorithms produce their results.4. Explanation: Systems and institutions that use algorithmic decisionmaking are encouraged to produce explanations regarding both the procedures followed by the algorithm and the specific decisions that are made. This is particularly important in public policy contexts.5. Data Provenance: A description of the way in which the training data was collected should be
maintained by the builders of the algorithms, accompanied by an exploration of the potential biases induced by the human or algorithmic data gathering process. Public scrutiny of the data provides maximum opportunity for corrections. However, concerns over privacy, protecting trade secrets, or revelation of analytics that might allow malicious actors to game the system can justify restricting access to qualified and authorized individuals.6. Auditability: Models, algorithms, data, and decisions should be recorded so that they can be audited in cases where harm is suspected.7. Validation and Testing: Institutions should use rigorous methods to validate their models and document those methods and results. In particular, they should routinely perform tests to assess and determine whether the model generates discriminatory harm. Institutions are encouraged to make the results of such tests public.
5. Allen Institute for Artificial Intelligence – AI2
Oren Etzioni, a professor of Computer Science and CEO of the Allen Institute for Artificial Intelligence has created three draft principles of AI Ethics. He first announced them in a New York Times Editorial, How to Regulate Artificial Intelligence (NYT, 9/1/17). See his TED Talk Artificial Intelligence will empower us, not exterminate us (TEDx Seattle; November 19, 2016). Etzioni says his proposed rules were inspired by Asimov’s three laws of robotics.
- An A.I. system must be subject to the full gamut of laws that apply to its human operator.
- An A.I. system must clearly disclose that it is not human.
- An A.I. system cannot retain or disclose confidential information without explicit approval from the source of that information.
We would certainly like to hear more. As Oren said in the editorial, he introduces these three “as a starting point for discussion. … it is clear that A.I. is coming. Society needs to get ready.” That is exactly what we are saying too. AI Ethics Work Should Begin Now.
Oren’s editorial included a story to illustrate the second rule on duty to disclose. It involved a teacher at Georgia Tech named Jill Watson. She served as a teaching assistant in an online course on artificial intelligence. The engineering students were all supposedly fooled for the entire semester course into thinking that Watson was a human. She was not. She was an AI. It is kind of hard to believe that smart tech students wouldn’t know that a teacher named Watson, who no one had ever seen or heard of before, wasn’t a bot. After all, it was a course on AI.
This story was confirmed by a later reply to this editorial by the Ashok Goel, the Georgia Tech Professor who so fooled his students. Professor Goel, who supposedly is a real flesh and blood teacher, assures us that his engineering students were all very positive to have been tricked in this way. Ashok’s defensive Letter to Editor said:
Mr. Etzioni characterized our experiment as an effort to “fool” students. The point of the experiment was to determine whether an A.I. agent could be indistinguishable from human teaching assistants on a limited task in a constrained environment. (It was.)
When we did tell the students about Jill, their response was uniformly positive.
We were aware of the ethical issues and obtained approval of Georgia Tech’s Institutional Review Board, the office responsible for making sure that experiments with human subjects meet high ethical standards.
Etzioni’s proposed second rule states: An A.I. system must clearly disclose that it is not human. We suggest that the word “system” be deleted as not adding much and the rule be adopted immediately. It is urgently needed not just to protect student guinea pigs, but all humans, especially those using social media. Many humans are being fooled every day by bots posing as real people and creating fake news to manipulate real people. The democratic process is already under siege by dictators exploiting this regulation gap. Kupferschmidt, Social media ‘bots’ tried to influence the U.S. election. Germany may be next (Science, Sept. 13, 2017); Segarra, Facebook and Twitter Bots Are Starting to Influence Our Politics, a New Study Warns (Fortune, June 20, 2017); Wu, Please Prove You’re Not a Robot (NYT July 15, 2017); Samuel C. Woolley and Douglas R. Guilbeault, Computational Propaganda in the United States of America: Manufacturing Consensus Online (Oxford, UK: Project on Computational Propaganda).
In the concluding section to the 2017 scholarly paper Computational Propaganda by Woolley (shown right) and Guilbeault, The Rise of Bots: Implications for Politics, Policy, and Method, they state:
The results of our quantitative analysis confirm that bots reached positions of measurable influence during the 2016 US election. … Altogether, these results deepen our qualitative perspective on the political power bots can enact during major political processes of global significance. …
Most concerning is the fact that companies and campaigners continue to conveniently undersell the effects of bots. … Bots infiltrated the core of the political discussion over Twitter, where they were capable of disseminating propaganda at mass-scale. … Several independent analyses show that bots supported Trump much more than Clinton, enabling him to more effectively set the agenda. Our qualitative report provides strong reasons to believe that Twitter was critical for Trump’s success. Taken altogether, our mixed methods approach points to the possibility that bots were a key player in allowing social media activity to influence the election in Trump’s favour. Our qualitative analysis situates these results in their broader political context, where it is unknown exactly who is responsible for bot manipulation – Russian hackers, rogue campaigners, everyday citizens, or some complex conspiracy among these potential actors.
Despite growing evidence concerning bot manipulation, the Federal Election Commission in the US showed no signs of recognizing that bots existed during the election. There needs to be, as a minimum, a conversation about developing policy regulations for bots, especially since a major reason why bots are able to thrive is because of laissez-faire API access to websites like Twitter. …
The report exposes one of the possible reasons why we have not seen greater action taken towards bots on behalf of companies: it puts their bottom line at risk. Several company representatives fear that notifying users of bot threats will deter people from using their services, given the growing ubiquity of bot threats and the nuisance such alerts would cause. … We hope that the empirical evidence in this working paper – provided through both qualitative and quantitative investigation – can help to raise awareness and support the expanding body of evidence needed to begin managing political bots and the rising culture of computational propaganda.
This is a serious issue that requires immediate action, if not voluntarily by social media providers, such as Facebook and Twitter, then by law. We cannot afford to have another election hijacked by secret AIs posing as real people.
As Etzioni stated in his editorial:
My rule would ensure that people know when a bot is impersonating someone. We have already seen, for example, @DeepDrumpf — a bot that humorously impersonated Donald Trump on Twitter. A.I. systems don’t just produce fake tweets; they also produce fake news videos. Researchers at the University of Washington recently released a fake video of former President Barack Obama in which he convincingly appeared to be speaking words that had been grafted onto video of him talking about something entirely different.
See: Langston, Lip-syncing Obama: New tools turn audio clips into realistic video (UW News, July 11, 2017). Here is the University of Washington YouTube video demonstrating their dangerous new technology. Seeing is no longer believing. Fraud is a crime and must be enforced as such. If the government will not do so for some reason, then self- regulations and individual legal actions may be necessary.
In the long term Oren’s first point about the application of laws is probably the most important of his three proposed rules: An A.I. system must be subject to the full gamut of laws that apply to its human operator. As mostly lawyers around here at this point, we strongly agree with this legal point. We also agree with his recommendation in the NYT Editorial:
Our common law should be amended so that we can’t claim that our A.I. system did something that we couldn’t understand or anticipate. Simply put, “My A.I. did it” should not excuse illegal behavior.
We think liability law will develop accordingly. In fact, we think the common law already provides for such vicarious liability. No need to amend. Clarify would be a better word. We are not really terribly concerned about that. We are more concerned with technology governors and behavioral restrictions, although a liability stick will be very helpful. We have a team membership openings now for experienced products liability lawyers and regulators.
6. Nicolas Economou
The latest attempt at articulating AI Ethics principles comes from Nicolas Economou, the CEO of electronic discovery search company, H5. Nicolas has a lot of experience with legal search using AI, as do several of us at AI-Ethics.com. In addition to his work with legal search and H5, Nicholas is involved in several AI ethics groups, including the AI Initiative of the Future Society at Harvard Kennedy School and the Law Committee of the IEEE’s Global Initiative for Ethical Considerations in AI.
Nicolas Economou has obviously been thinking about AI ethics for some time. He provides a solid scientific, legal perspective based on his many years of supporting lawyers and law firms with advanced legal search. Economou has developed six principles as reported in an ABA Legal Rebels article dated October 3, 2017, A ‘principled’ artificial intelligence could improve justice. (Some of the explanations have been edited out as indicated below. Readers are encouraged to consult the full article.) As you can see the explanations given here were written for consumption by lawyers and pertain to e-discovery. They show the application of the principles in legal search. See eg TARcourse.com. The principles have obvious applications in all aspects of society, not just the Law and predictive coding, so their value goes beyond the legal applications here mentioned.
Principle 1: AI should advance the well-being of humanity, its societies, and its natural environment. The pursuit of well-being may seem a self-evident aspiration, but it is a foundational principle of particular importance given the growing prevalence, power and risks of misuse of AI and hybrid intelligence systems. In rendering the central fact-finding mission of the legal process more effective and efficient, expertly designed and executed hybrid intelligence processes can reduce errors in the determination of guilt or innocence, accelerate the resolution of disputes, and provide access to justice to parties who would otherwise lack the financial wherewithal.
Principle 2: AI should be transparent. Transparency is the ability to trace cause and effect in the decision-making pathways of algorithms and, in hybrid intelligence systems, of their operators. In discovery, for example, this may extend to the choices made in the selection of data used to train predictive coding software, of the choice of experts retained to design and execute the automated review process, or of the quality-assurance protocols utilized to affirm accuracy. …
Principle 3: Manufacturers and operators of AI should be accountable. Accountability means the ability to assign responsibility for the effects caused by AI or its operators. Courts have the ability to take corrective action or to sanction parties that deliberately use AI in a way that defeats, or places at risk, the fact-finding mission it is supposed to serve.
Principle 4: AI’s effectiveness should be measurable in the real-world applications for which it is intended. Measurability means the ability for both expert users and the ordinary citizen to gauge concretely whether AI or hybrid intelligence systems are meeting their objectives. …
Principle 5: Operators of AI systems should have appropriate competencies. None of us will get hurt if Netflix’s algorithm recommends the wrong dramedy on a Saturday evening. But when our health, our rights, our lives or our liberty depend on hybrid intelligence, such systems should be designed, executed and measured by professionals with the requisite expertise. …
Principle 6: The norms of delegation of decisions to AI systems should be codified through thoughtful, inclusive dialogue with civil society. … The societal dialogue relating to the use of AI in electronic discovery would benefit from being even more inclusive, with more forums seeking the active participation of political scientists, sociologists, philosophers and representative groups of ordinary citizens. Even so, the realm of electronic discovery sets a hopeful example of how an inclusive dialogue can lead to broad consensus in ensuring the beneficial use of AI systems in a vital societal function.
Nicolas Economou believes, as we do, that an interdisciplinary approach, which has been employed successfully in e-discovery, is also the way to go for AI ethics. Note his use of the word “dialogue” and mention in the article of The Sedona Conference, which pioneered the use of this technique in legal education. We also believe in the power of dialogue and have seen it in action in multiple fields. See eg. the work of physicist, David Bohm and philosopher, Martin Buber. That is one reason that we propose the use of dialogue in future conferences on AI ethics. See the AI-Ethics.com Mission Statement.