{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"# A lesson in AI ethics  \n\n## Introduction \n\n\n Imagine you have programmed a robot, and its purpose is to find the cause of climate change, find the solution, and end the climate crisis. You gave the robot the command and your robot obeyed it. Soon your robot identifies the problem, then finds the solution and now starts to implement its solution. When you look at the data the robot collected, to your horror, the robot has concluded that humans are causing the climate crisis and its solution is to destroy humans. This is not what you hoped the outcome would be but the robot is only obeying your commands. AI is evolving at a rapid pace ushering in a new era of advancements and knowledge in many industries. The question is: what mechanisms can we keep in place to prevent unforeseeable consequences in AI like in the scenario above.\n![image](https://emtemp.gcom.cloud/ngw/globalassets/en/newsroom/images/graphs/AI%20Press%20release.png)\n\nThe estimated mean of AI or ML projects per year for each organization.\n\nCredit: from the article “Gartner Survey Reveals Leading Organizations Expect to Double the Number of AI Projects In Place Within the Next Year”\n\n![image](https://virtualizationreview.com/articles/2023/04/05/~/media/ECG/virtualizationreview/Images/2023/04/ai_incidents_s.ashx)\n\nThe number of AI misuse from 2012-2021\n\nCredit: From Stanford's AI Index Report \n\nIf you compare the two graphs from above, as more projects in AI are being created the number of misuse in AI goes up.   In order to prevent the misuse of AI as AI continues to grow, we now need ethical ideas in practice. This is where AI ethics comes in.\n\n\n## What is AI ethics?\n\n\nAI ethics are the moral considerations to the use and creation of AI with its possible outcomes. These considerations include safety, privacy, a non-bias system, human concerns, managing AI’s environmental impact, preventing AI mistakes, and morally right and wrong things that we as a society agree upon. In addition to this AI ethics also address the responsibility of the creators and consumers of AI systems. Many businesses and organizations have their own guidelines when it comes to the use of AI since each business or organization has their own values and not all AI systems have the same purpose. Not all AI systems may share the same purpose or use but generally speaking most of these Ai systems share key principles in order to meet the moral considerations of our society today. \n\n## AI Key principals\n\n\nBefore the world of AI, the famous science fiction writer Isaac Asimov realized the potential dangers of AI long before its creation. Due to this reason Isaac Asimov created the three laws of robotics to prevent AI from being misused or becoming too powerful. In Asimov’s three laws of robotics the first law states that a robot cannot hurt a human or allow a human to come to harm. The second law states that robots must obey humans not unless it conflicts with the first law.  The third law states that robots must protect themselves (since they are very expensive) not unless it conflicts with the first and second laws. Imagine the world without Asimov’s three laws of robotics or if the laws were left out of order.\n\n\n\n","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19"}},{"cell_type":"markdown","source":"![Image](https://imgs.xkcd.com/comics/the_three_laws_of_robotics.png)\n\nCredit: From the website https://xkcd.com/1613/ ","metadata":{}},{"cell_type":"markdown","source":"Even though Asimov’s three laws of robotics served as a good basis for early AI systems, AI systems have become more complex. Asimov’s three laws of robotics are no longer enough. In recent years there have been conference’s to address what morals, principles, and considerations should be put in place in regards to the development of AI. Some examples of these principles that have been created in recent years are…\n\n* On November 25 2021 UNESCO (United Nations Educational, Scientific and Cultural Organization) created its first ever global AI principles handbook titled “Recommendation on the Ethics of Artificial Intelligence.” Over 193 members have adopted these AI principles. \n\n* On June 28 2021 the WHO (World Health Organization) came out with its own AI principles handbook titled “Ethics & Governance of Artificial Intelligence for Health” This handbook is meant to be a guide for how AI should be used in the health industry. \n\n* On June 28, 2023 the Vatican released its own handbook guide titled “Ethics in the Age of Disruptive Technologies: An Operational Roadmap.” In this handbook it has seven AI principles and 43 actionable steps that many tech companies can incorporate into their company. This hand book was created in collaboration with Santa Clara University’s Markkula Center for Applied Ethics. \n\nOf course more AI principles have been created, however the most notable example of AI principles are Asilomar principles. In 2017, the Future of Life Institute hosted the Beneficial AI 2017 conference, at the Asilomar Conference Grounds in California to discuss ethical guidelines that could be left in place in the field of AI. During this conference, researchers broke into groups to discuss principles from the list created prior to the meeting that stated how technology should be handled. After reviewing this list and giving suggestions on how it can be improved, the considerations given created new principles or improved versions of the principles presented in the list of several versions of principles competing against each other. Finally after surveying the attendees at the conference to figure out the level of support for all principles presented at the meeting, the Asilomar principles were created. Asilomar principles consist of three main topics: research, ethics and values, and longer term issues. In total there are 23 guidelines and here there are as follows: \n\n\n\n### Research Issues\n1) Research Goal: The goal of AI research should be to create not undirected intelligence, but beneficial intelligence.\n\n2) 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\n\n3) Science-Policy Link: There should be constructive and healthy exchange between AI researchers and policy-makers.\n\n4) Research Culture: A culture of cooperation, trust, and transparency should be fostered among researchers and developers of AI.\n\n5) Race Avoidance: Teams developing AI systems should actively cooperate to avoid corner-cutting on safety standards.\n\n### Ethics and Values\n\n6) Safety: AI systems should be safe and secure throughout their operational lifetime, and verifiably so where applicable and feasible.\n\n7) Failure Transparency: If an AI system causes harm, it should be possible to ascertain why.\n\n8) Judicial Transparency: Any involvement by an autonomous system in judicial decision-making should provide a satisfactory explanation auditable by a competent human authority.\n\n9) 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.\n\n10) 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.\n\n11) Human Values: AI systems should be designed and operated so as to be compatible with ideals of human dignity, rights, freedoms, and cultural diversity.\n\n12) 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.\n\n13) Liberty and Privacy: The application of AI to personal data must not unreasonably curtail people’s real or perceived liberty.\n\n14) Shared Benefit: AI technologies should benefit and empower as many people as possible.\n\n15) Shared Prosperity: The economic prosperity created by AI should be shared broadly, to benefit all of humanity.\n\n16) Human Control: Humans should choose how and whether to delegate decisions to AI systems, to accomplish human-chosen objectives.\n\n17) 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.\n\n18) AI Arms Race: An arms race in lethal autonomous weapons should be avoided.\n\n### Longer-term Issues\n\n19) Capability Caution: There being no consensus, we should avoid strong assumptions regarding upper limits on future AI capabilities.\n\n20) 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.\n\n21) Risks: Risks posed by AI systems, especially catastrophic or existential risks, must be subject to planning and mitigation efforts commensurate with their expected impact.\n\n22) 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.\n\n23) 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.\n","metadata":{}},{"cell_type":"markdown","source":"![image 2](https://futureoflife.org/wp-content/uploads/2017/01/principled_conversation.jpg)\n\nA picture taken from the Asilomar AI conference\n\nCredit: From the Future of Life Institute website","metadata":{}},{"cell_type":"markdown","source":"### Flaws\n\n\nThe Asilomar principle has received over 90% of support from the attendees that attended the conference and has received over 5,000 signatures in support. Some notable signatures are Elon Musk and the late Stephen Hawking.  Even though there is wide support for Asilomar principles there still some flaws:\n\n\n* Asilomar principles are too vague which makes it ineffective since different people can interpret it differently but it can be applied in many contexts.\n* Asilomar principles are not the law which means companies do not have to follow Asilomar principles.\n* Asilomar principles assume the creators and consumers of AI do not have any bad intentions and this is not true since there are some people out there that want to misuse AI.\n* The principles have not mentioned any policies of AI taking over jobs or its socioeconomic impact since Asilomar principles only focus on creation and use of AI. \n* There is no focus on conflicting principles. An example of this is using data to train the AI model but this could conflict with the principle of privacy.\n\nThese flaws do not only apply to Asilomar principles but these concerns apply to the many guidelines out there that address AI ethics. Finally a concern to note about principles that address AI ethics is that even though they are guidelines for developers and users of AI these guidelines are often not applied in government policy. \n\n## Actionable AI ethics\n\n\nIn recent years there has been an attempt to bridge AI principles and AI government policy. The only challenge here is that AI principles are not directed at government policy making. Actionable AI ethics addresses the fact on how to make AI principles into actionable items in government policy. When talking about Actionable AI ethics we will be looking at these four main points:\n\n\n1. Multidisciplinary research\n2. preliminary landscape assessments\n3. Maintaining human control in AI systems\n4. Societies' role in AI ethics.\n\n\n\n### Multidisciplinary research \n\nIn the research community, there is an agreed notation that there is a need for multidisciplinary research. Multidisciplinary research helps us see the advantages and the consequences of an AI system to help us better understand its benefit to society, and help us prevent some of the dangers. In addition, the current reality we live in does not live up to ideals we see in AI ethics. This is a problem since AI is data driven and the data is at risk of being biased because of the society we live in. An example of biases in AI systems is in 2018, it was discovered that Amazon's hiring system favored men over women. This happened because in order to train Amazon’s AI hiring system, they used applicants' resumes from the past ten years. The AI system noticed a pattern of how men were preferred over women, and this pattern was reflected in Amazon’s hiring system. There is still a huge disparity when it comes to women working in tech companies.\n\n![image](data:image/png;base64,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)\n\nCredit: From the article “Amazon scraps secret AI recruiting tool that showed bias against women”\n\nThis emphasizes the importance of needing many different expertise in multidisciplinary research in order to improve the AI ethics we have today and to see AI’s consequences in the long- term. \n\n\n### Preliminary landscape assessment \n\nIn the article “Three Pathways for Actionable Principles for AI Policy” it talks about three things needed in order to make AI ethics actionable. One of the things it addresses is the need for a preliminary landscape assessment. A landscape assessment can help us see weak spots or some concerns before they become a problem. In addition to this it can inform us of an AI’s system stages in development. Things needed to do a landscape assessment are identifying the issue, identifying key stakeholders, potential ethical implications (like fairness, transparency and accountability), and the practical case. Landscape assessments can make actionable AI more impactful and applicable by giving us insight into the creation of actionable AI ethics. \n\n\n\n### Maintaining human control in AI systems\n\nWhen we are leaving AI systems in human environments, there has been talk about needing human control over these AI systems. The reason behind this is because when we leave AI systems in human environments, there could be some unforeseen consequences. In addition, AI’s rapid growth could lead to clashes with societal norms and human values. Also, if undesired things happen, who would be responsible? The first step in order to address this is to be able to have tracking and tracing within these human-AI systems. The tracking concept addresses the fact that a human- AI system under any circumstance should be responsive to human moral reasoning. The tracing concept states that everything within the human-AI system should be traceable to one human who is part of the human-AI system or responsible for its creation. These two principles serve as a good foundation for maintaining humans in AI systems. The problem is that it is very hard in order to put these concepts into practice. In order to address this here are a couple of more points we need to consider: \n\n\n\n* The first thing we need to consider is making sure the human-AI system stays within its moral operational design. \n\n* The second thing is that within the human-AI system is that there should be a model that can explain the human-AI system. These models are called mental models.  Here is an example:","metadata":{}},{"cell_type":"markdown","source":"![image 3](https://www.researchgate.net/publication/350512179/figure/fig1/AS:1007261824585728@1617161341251/This-framework-displays-the-player-AI-interaction-and-players-mental-model-process-of-1.jpg)\n\n“This framework displays the player-AI interaction and player's mental model process of 1) attributing failure, 2) discovering the AI's capabilities, and 3) strategizing with this knowledge in the game.”\n\nCredit: From the paper “Understanding Mental Models of AI through Player-AI Interaction” \n\n","metadata":{}},{"cell_type":"markdown","source":"\n* The third thing is that it should be clear what the humans and AI agents' roles are in the human AI-system in order to create accountability in case something goes wrong. An example of this is an autonomous vehicle. A human can take control of the autonomous vehicle anytime but if the driver loses awareness of the situation they are in or if they lack the skill to take control since they barely engage within the system this will limit the drivers ability to have full control of the system. This would mean the human is not fully responsible and there could have been something that went wrong with the AI system as well. \n\n\n* The final thing is that within the human-AI system, whatever the actions of the system are, it should be traceable to the humans involved with their moral understanding. \n\n\n### Societies' role in AI ethics.\n\nSomething almost always left aside when talking about actionable AI is how society can play a role in the advancement of AI ethics. In order to bring society into the conservation of Actionable AI ethics is by focusing on human rights values like freedom, peace, justice, etc when talking about actionable AI. The reasoning behind this is that if we focus on human rights values, it is easier for society to understand their role in the development of AI and it is easier for law enforcement to get involved. Active roles members of society can play roles in actionable AI is by holding companies accountable for how they use AI and also making sure that they hold themselves for how they use AI as well. Finally, the most important thing society can do is to educate themselves on AI so that they will have a better understanding of how AI is going to impact their communities. \n\n\nNow that we have finished talking about the main four topics needed to talk about when talking about actionable AI ethics, let's see how AI ethics plays out in the real world.\n\n\n\n## AI ethics in the real world\n\n\nSo far in this essay, we have covered the basis about AI ethics but now let’s see how it all works together in the real world. In this section we will look at AI advancements in the past two years in the health care, creative, and arms industries. \n\n\n\n### Health care industry\n\nThe most notable advancement of AI in the healthcare industry is the fact that AI can make accurate diagnosis of many diseases within a matter of seconds and more accurately than most doctors. An example of this is using AI to diagnose patients with prostate cancer. In an article titled “Can Artificial Intelligence Help See Cancer in New, and Better Ways?” published on March 22, 2022, Ismail Baris Turkbey, M.D a radiologist was asked to outline where prostate cancer is in the brain image on one side of the screen. On the other side of the screen an AI model was asked to do the same task. When comparing both of these results it shows that the AI model and the radiologist had produced similar results.\n\n","metadata":{}},{"cell_type":"markdown","source":"![Image 4](https://www.cancer.gov/sites/g/files/xnrzdm211/files/styles/cgov_article/public/cgov_image/media_image/2022-03/mpMRI%20prostate%20AI.jpg?h=85ef9b55&itok=AJ5_5xmo)\n\n“On a multiparametric MRI scan of a patient's prostate, a cancer-suspicious area (red) is highlighted by an AI model developed by Dr. Turkbey.”\n\nCredit : Stephannie Harmon , Ph.D and from the article “Can Artificial Intelligence Help See Cancer in New, and Better, Ways?\n","metadata":{}},{"cell_type":"markdown","source":"Despite the promise of AI in the medical field there are a few ethical problems that need to addressed: \n\n\n* The black box problem: Many AI models used in the medical fields use DL (deep learning) to diagnose different diseases. The main problem with using DL is that it does not explain how it arrived at its outputs from the inputs that were given, because of this it has been referred to as a “black box.” \n\n* Use of data: In order to train the AI models it uses huge amounts of data. The only problem with this is that it could invade a patient's privacy. \n\n\n\n\n### Creative industry \n\nThe creative industry has made many advancements when it comes to the media, arts and social media. An example of this is when ChatGPT came into existence on November 30, 2022. ChatGPT which has been used to create human-like conservations. In addition to this it has been used to help people with their essays, emails and codes. \n\n","metadata":{}},{"cell_type":"markdown","source":"![image 5](https://assets.website-files.com/6344c9cef89d6f2270a38908/64148ed756708f9b82464c96_image-of-hand-holding-an-ai-face-looking-at-the-words-chatgpt-openai.webp)\n\nCredit: From the article “What Is Chat GPT: Basics For Product Development In 2023”","metadata":{}},{"cell_type":"markdown","source":"Despite how many people ChatGPT has helped there are some ethical concerns worth notating:\n\n* Biases: ChatGPT has been trained on numerous data but some of that data may have some biases due to the society we live in. This can prove to be a problem when ChatGPT is asked to write essays because some of those biases may be present within the essay or in the things it is asked to produce. \n\n\n* Copyright infringement: there have been many concerns with ChatGPT copying other texts on the internet.\n\n* Privacy violations: there have been some concerns that ChatGPT is using people’s personal information which violates someone’s privacy. \n\n\n\n### Arms industry \n\nThe arms industry in other words the defense industry is making many advancements when it comes to AI in defense, weapons, drones, etc. One notable advancement are autonomous weapons or known as “killer bots.” Autonomous weapons are weapons that don’t have any human intervention and can select targets and strike them. The use of these autonomous weapons have many ethical challenges that need to be addressed:\n\n\n* Human intervention: There is no human intervention in these autonomous weapons that are designed to kill people. Without human intervention there would be no morals to guide these autonomous weapons. \n\n* Responsibility: If humans were to harm civilians then they can be held accountable because whoever hurt them would be the one who would be held accountable. However, with autonomous weapons since there is no human intervention it can be very hard to hold them accountable. \n\n","metadata":{}},{"cell_type":"markdown","source":"![image 6](https://lop.parl.ca/staticfiles/PublicWebsite/assets/images/ResearchPublications/InBriefs/2019-55-fig1-e.png)\n\nThe above picture shows how an non-autonomous weapon system works vs. an autonomous weapon system works. \n\nCredits:  From the article “Autonomous Weapon Systems: Selected Implications for International Security and for Canada” and from the Library of Parliament based on information obtained from Paul Scharre, Army of None: Autonomous Weapons and the Future of War, W.W. Norton and Company, New York, 2018. Illustration of the fully autonomous weapon system based on IAI’s HARPY.","metadata":{}},{"cell_type":"markdown","source":"## The importance of AI ethics\n\nThroughout this essay we have talked so much about AI ethics and advancements in AI. The one thing we haven’t addressed is why AI ethics are so important. AI ethics is important because it helps guide the development of AI in the right direction. Without any rules in the development of AI it can easily be misused and harm our society. Everything in this world has rules that we need to follow in order to keep order. AI is no exception and as AI advances we would also always have to make sure there are ethical guidelines that are being put in place with the new technologies being created to make sure AI is being used to its full potential without being misused. \n\n\n\n## Conclusion \n\nIn this paper we have talked about \n\n* What is AI ethics\n* Key principles of AI ethics\n* Actionable AI ethics\n* AI ethics in the real world \n* The importance of AI ethics\n\n\n\nThe future of AI is still really unknown as we still don’t know AI’s full capabilities. As we are uncovering more of what AI is truly capable of, it is important to always implement mechanisms to prevent AI from becoming too powerful. Yes the future of AI seems very scary with concerns of robots taking over the world or putting an end to humanity. It is also important to note that AI can benefit society greatly if used correctly since it has already benefited humanity so much. We should not be afraid to explore the world of AI as AI is the future. As AI is developing, it is important to make sure that there are ethical guidelines in place with the new technologies that are being created. “The potential benefits of artificial intelligence are huge, so are the dangers.” ~Dave Waters. \n\n\n\n\n## References \n\n1. A guide to artificial intelligence in the enterprise https://www.techtarget.com/whatis/definition/AI-code-of-ethics\n\n2. AI Ethics: What It Is And Why It Matters\nhttps://www.forbes.com/sites/nishatalagala/2022/05/31/ai-ethics-what-it-is-and-why-it-matters/?sh=71d0ac3e3537\n\n3. AI principles \nhttps://futureoflife.org/open-letter/ai-principles/\n\n4. Exploring the Asilomar AI Principles: A Guide to Ensuring Safe and Beneficial AI Development https://www.techopedia.com/exploring-the-asilomar-ai-principles-a-guide-to-ensuring-safe-and-beneficial-ai-development#:~:text=The%20principles%20address%20 concerns%20about,all%20humans%20and%20the%20environment\n\n5. Actionable Principles for Artificial Intelligence Policy: Three Pathways https://link.springer.com/article/10.1007/s11948-020-00277-3\n\n6. Meaningful human control: actionable properties for AI system development https://link.springer.com/article/10.1007/s43681-022-00167-3\n\n7. The Quest for Actionable AI Ethics (draft) https://repository.up.ac.za/bitstream/handle/2263/79676/RuttkampBloem_Quest_2020.pdf?sequence=1&isAllowed=y\n\n8. Can Artificial Intelligence Help See Cancer in New, and Better, Ways? https://www.cancer.gov/news-events/cancer-currents-blog/2022/artificial-intelligence-cancer-imaging\n\n9. Artificial intelligence for assisting cancer diagnosis and treatment in the era of precision medicine \nhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC8626610/\n\n10. Ethical implications of ChatGPT: The Good, The Bad, The Ugly\nhttps://unstop.com/blog/ethical-implications-of-chatgpt\n\n11. What is ChatGPT and why does it matter? Here's what you need to know\nhttps://www.zdnet.com/article/what-is-chatgpt-and-why-does-it-matter-heres-everything-you-need-to-know/\n\n12. What you need to know about autonomous weapons\nhttps://www.icrc.org/en/document/what-you-need-know-about-autonomous-weapons\n\n13. Peter Maurer: \"Autonomous weapon systems raise ethical concerns for society\" https://reliefweb.int/report/world/peter-maurer-autonomous-weapon-systems-raise-ethical-concerns-society \n\n14. THE IMPORTANCE OF ETHICS IN AI \nhttps://www.wizata.com/knowledge-base/the-importance-of-ethics-in-ai\n\n15. Embracing Actionable Principles of AI Implementation: Pathways, Properties, and Human-Centric Approaches \nhttps://www.kaggle.com/code/luckypen/actionable-ai-ethics \n\n16. Amazon scrapes secret project \nhttps://www.reuters.com/article/us-amazon-com-jobs-automation-insight-idUSKCN1MK08G\n\n17. WHO issues first global report on Artificial Intelligence (AI) in health and six guiding principles for its design and use \nhttps://www.who.int/news/item/28-06-2021-who-issues-first-global-report-on-ai-in-health-and-six-guiding-principles-for-its-design-and-use\n\n18. Ethics and governance of artificial intelligence for health https://www.who.int/publications/i/item/9789240029200\n\n19. Ethics of Artificial Intelligence\nhttps://www.unesco.org/en/artificial-intelligence/recommendation-ethics\n\n20. The Vatican creates own guidelines against misuse of AI in ChatGPT era https://www.sakshipost.com/news/vatican-creates-own-guidelines-against-misuse-ai-chatgpt-era-204084\n\n21. The uselessness of AI ethics \nhttps://link.springer.com/article/10.1007/s43681-022-00209-w\n\n22. 2023 Kaggle AI Report: AI Ethics https://www.kaggle.com/code/raimondextervinluan/2023-kaggle-ai-report-ai-ethics/notebook#PART-1:-NEWS\n\n\n\n\n","metadata":{}}]}