\n[$\\mathit{1.\\ Introduction\\ and\\ Background}$](#2)
\n[$\\mathit{2.\\ Recent\\ Issues:\\ Privacy,\\ Accountability,\\ Transparency\\ on\\ AI}$](#3)
\n[$\\mathit{3.\\ Challenges:\\ Misconception,\\ Generative\\ AI\\ Models,\\ and\\ Collaborative\\ Structure\\ for\\ the\\ AI\\ Ethics\\ }$](#4)
\n[$\\mathit{4.\\ Discussions:\\ Metrics,\\ Quality,\\ Lessons\\ Learned\\ }$](#5)
\n[$\\mathit{5.\\ Conclusions}$](#6)
\n[$\\mathit{Reference}$](#9)
","metadata":{"_uuid":"91993261-63c9-4368-9a92-11dd00e905c4","_cell_guid":"a8eaed51-d18c-4926-9ef4-f594781b0855","id":"SJW_4UpQxohG","trusted":true}},{"cell_type":"markdown","source":"\n\n\n\n\n\n
Abstract
\n\nThe field of AI ethics holds immense importance in comprehending and tackling the ethical implications caused by artificial intelligence (AI) in the proliferating AI technology. Its core focus lies in integrating ethical principles into AI systems to foster responsible practices and address potential risks. By embracing ethical considerations, we can ensure the protection of privacy, promote fairness, and enhance societal harmony. To effectively govern AI systems, robust regulations and policies are mandatory. Transparent and diverse frameworks that prioritize accountability and safeguard user rights are crucial for establishing a strong and safe foundation. This article acknowledges the multifaceted challenges posed by issues such as biases, job displacement, privacy breaches, and the potential misuse of AI in domains like cyber warfare and propaganda, hate speeches, deepfakes and the list goes on.\n\nThis article underscores the significance of collaborative efforts among policymakers, researchers, and industry practitioners. It emphasizes the need for interdisciplinary perspectives, active stakeholder involvement, and continuous research and education to foster a culture of responsible AI development and commercialization. AI ethics strives to inspire proactive measures that prioritize the ethical and responsible use of AI technologies. It highlights the importance of effective governance and regulatory frameworks that strike a balance between innovation and ethical considerations. Ultimately, the objective is to harness the vast potential of AI while ensuring it serves the best interests of humanity and acts as an aid to mankind not a bane to humanity.\n
\n\n\n","metadata":{"_uuid":"31ad8d18-37a5-43c0-a72e-123958619238","_cell_guid":"58a70ff7-5df8-421c-93f5-3e70f719c729","id":"NmoxtjiCxohG","trusted":true}},{"cell_type":"markdown","source":"\n\n\n\n\n\n1. Introduction
\nMotivation
\n \nArtificial Intelligence (AI) has emerged as a transformative force, revolutionizing various industries and impacting our daily lives. However, with the rapid advancements in AI, addressing the ethical implications surrounding its development and deployment is essential [1]. AI ethics provides a critical framework to guide responsible practices, ensuring that AI systems align with societal values and respect fundamental rights. This essay delves into the need for AI ethics in the current scenario, examines the steps taken thus far, discusses the hurdles encountered, clarifies the distinction between unethical practices and innovation, explores what more needs to be done, emphasizes the core values of AI ethics, highlights its role in human and technological advancements, and outlines the future scope of this crucial field.\n\n The history of Artificial Intelligence has started way back before the 21st century, long before the “disagreements” between the countries that led to World War. In the latter War era (World War II), the advancement of Computer Intelligence or known as “Machine Intelligence” had just laid its one of the foundations through decrypting codes to decode the Enigma [cite here]. Enigma had become one of the prominent techniques that only desired party included inside can clearly understand the received messages or information. Within this machine lies down infinite possibilities to be deciphered, which was almost impossible to crack down or tagged as “unbreakable” machine. However with such efforts from mathematicians that time including the famous Alan Turing, the encoded message from the machine was deciphered and simplified quintillions of possibilities for decrypting the message. We simply can’t dispose that the development Artificial Intelligence are contributed from such method - pattern comparison for such encrypted message to be predicted and decoded, somewhat similar to the Machine Learning’s Black Box settings.Background
\n
\n Turing’s contributions (infamous machine namely Bombes) has put our mind into the similarity of what AI actually is, there are red lines where Artificial Intelligence are influenced by Turing’s concept of a “Universal Computing Machine” or known as “Turing Machine”. This idea comprehensively put understanding towards how Computation works and the ability of performing complex calculations and tasks. It yields such properties where “black boxes” are included and it’s not easily explainable, it required sum amount of efforts to interpret and understand the decision making processes. Hence, giving us the big picture where ‘Systems can process information, learns, make predictions and decisions based on patterns’.
\n In the current scenario, AI technologies are expediting prominence and integration across various sectors such as healthcare, finance, and transportation. Consequently, ethical concerns related to privacy, transparency, bias, accountability, and potential unintended consequences have arisen. The need for AI ethics has become paramount to ensure responsible development, deployment, and governance of AI systems that are in line with societal values and respect individual rights.\n
Figure 1: 3 The history of AI.
2. Recent Issues
\n\n On top of Artificial Intelligence itself, implementing AI Ethics faces several hurdles. The lack of standardized ethical guidelines poses challenges in ensuring consistent ethical practices across different AI systems and applications. Additionally, the complexity of AI algorithms and the potential for unintended biases make it difficult to address fairness and transparency concerns. Ethical dilemmas arise when striking a balance between innovation and regulation, necessitating careful consideration and interdisciplinary collaboration. Thus, we addressed there are key issues in the AI Ethics: Privacy, Transparency, and Accountability.\n As an example for the AI ethics that faced an issue, where in the news [5-8] current tools on AI, namely ChatGPT has been facing privacy leakage issue since it is disclosing some personal information submitted by the user. It gives clear message that security and privacy are very prone to the exploitation of unwanted parties. Nowadays the circulation of the information are pretty fast and published in such enormous amount, which enhances people to accidentally leak their own private information in such way: clicking ads, unknown links, Accept Cookies, and many more. It might seem that these mentioned actions are considered minor and do no harm to their ends, but it lies a certain risk where often people losing their valuable information, and to some extent yield the probability of risking their life.\n In the current world where we sits nicely in, there are around seven millions communications tower or Base Stations, and more than six billions mobile terminals in the world and these numbers are nowhere to be stopped in foreseeable future [9,10]. It is quite the heavy-loaded tasks to collect all information and data from these different infrastructures and passing it around through different services without having a privacy leakage, let alone its computing complexity and energy consumption.\n \n The challenges are not easy for AI, we also like to address some main challenges faced by the AI Ethics. The Terms \"Artificial Intelligence\" has faced so many misconception, with the current uprising topic under the hood of Generative AI Models, it is very very important to address the urgency and needs so that AI can be massively adopted in many domains, without worrying the beforementioend issues in the previous chapters \n \n Given the AI emergence, there has been a lot of usage in these terms, but for real what is the terms \"Artificial Intelligence\"? Is it merely just predicting or recognizing some text, images, patterns? Hence, you call them Artificial Intelligence? Based on what you have read so far, there would be a lot interpretation about Artificial Intelligence, but simply putting it down, the foundations are: Mathematics. Do you think Probabilistic and Statistic are on their own branch? Do you think the K-Mean Classifier are on their own end as well? Do you think the data acquired and no processing then the machine can learn by itself without we define the classifier, or learning from the 'clustered' end or the distribution itself? NO, to name a few, Maximum Likelihood also one of the techniques to learn. But simply putting the Foundation of it all: Mathematics Fundamental are in play. \n\n \n Given the AI emergence models. It composes of 2 components, a encoder and a decoder, which are both multi-layer neural network. Given some samples $X$, which follow a distribution $p(X)$. Instead of learning $p(X)$ directly, VAE introduces a laten variable $Z$ to learn the distribution by the formula $p(X)=\\sum\\limits_{z} p(X|Z)p(Z)$. $p(X|Z)$ is called Posterior Distribution and $p(Z)$ is called Prior Dsitribution. However, this is very resource-consuming, since we need to sample $Z$ many times to estimate $p(X)$. Consequently, another distribution $p(Z|X)$ is introduced, from which $Z$ is sampled, making it easier to estimate $p(X)$. \n \n \n \n Given the AI emergence models. It composes of 2 components, a encoder and a decoder, which are both multi-layer neural network. Given some samples $X$, which follow a distribution $p(X)$. Instead of learning $p(X)$ directly, VAE introduces a laten variable $Z$ to learn the distribution by the formula $p(X)=\\sum\\limits_{z} p(X|Z)p(Z)$. $p(X|Z)$ is called Posterior Distribution and $p(Z)$ is called Prior Dsitribution. However, this is very resource-consuming, since we need to sample $Z$ many times to estimate $p(X)$. Consequently, another distribution $p(Z|X)$ is introduced, from which $Z$ is sampled, making it easier to estimate $p(X)$. \n\n\n\n \n\n","metadata":{"_uuid":"4312017b-8055-46e3-b927-0cd137156703","_cell_guid":"821411f7-96e6-47ff-9e58-8e38cf508e41","id":"wwINefP4xohI","trusted":true}},{"cell_type":"markdown","source":"","metadata":{"_uuid":"ca1551e8-117c-4853-9598-6167bb8b14c8","_cell_guid":"2eb7784f-5f24-4220-8e8b-e764e57c6ec0","id":"mzQfmR18yVTb","trusted":true}},{"cell_type":"markdown","source":"\n The challenges are not easy for AI, we also like to address some main challenges faced by the AI Ethics. The Terms \"Artificial Intelligence\" has faced so many misconception, with the current uprising topic under the hood of Generative AI Models, it is very very important to address the urgency and needs so that AI can be massively adopted in many domains, without worrying the beforementioend issues in the previous chapters \n \n Given the AI emergence, there has been a lot of usage in these terms, but for real what is the terms \"Artificial Intelligence\"? Is it merely just predicting or recognizing some text, images, patterns? Hence, you call them Artificial Intelligence? Based on what you have read so far, there would be a lot interpretation about Artificial Intelligence, but simply putting it down, the foundations are: Mathematics. Do you think Probabilistic and Statistic are on their own branch? Do you think the K-Mean Classifier are on their own end as well? Do you think the data acquired and no processing then the machine can learn by itself without we define the classifier, or learning from the 'clustered' end or the distribution itself? NO, to name a few, Maximum Likelihood also one of the techniques to learn. But simply putting the Foundation of it all: Mathematics Fundamental are in play. \n\n \n Given the AI emergence models. It composes of 2 components, a encoder and a decoder, which are both multi-layer neural network. Given some samples $X$, which follow a distribution $p(X)$. Instead of learning $p(X)$ directly, VAE introduces a laten variable $Z$ to learn the distribution by the formula $p(X)=\\sum\\limits_{z} p(X|Z)p(Z)$. $p(X|Z)$ is called Posterior Distribution and $p(Z)$ is called Prior Dsitribution. However, this is very resource-consuming, since we need to sample $Z$ many times to estimate $p(X)$. Consequently, another distribution $p(Z|X)$ is introduced, from which $Z$ is sampled, making it easier to estimate $p(X)$. \n \n \n \n Given the AI emergence models. It composes of 2 components, a encoder and a decoder, which are both multi-layer neural network. Given some samples $X$, which follow a distribution $p(X)$. Instead of learning $p(X)$ directly, VAE introduces a laten variable $Z$ to learn the distribution by the formula $p(X)=\\sum\\limits_{z} p(X|Z)p(Z)$. $p(X|Z)$ is called Posterior Distribution and $p(Z)$ is called Prior Dsitribution. However, this is very resource-consuming, since we need to sample $Z$ many times to estimate $p(X)$. Consequently, another distribution $p(Z|X)$ is introduced, from which $Z$ is sampled, making it easier to estimate $p(X)$. \n\n\n\n \n\n","metadata":{"_uuid":"64d81266-3189-40b8-b98d-836fcc645ea8","_cell_guid":"e9936472-fcac-4ffa-801a-87cc3934fb24","id":"WgMoW31oyVsE","trusted":true}},{"cell_type":"markdown","source":"\n In conclusion, the growing. \n \n\n","metadata":{"_uuid":"48c31762-042a-419b-bcfd-9d382fcd3b2d","_cell_guid":"9f1b1537-4262-4427-aa41-2ce5505e4c07","id":"O8lY6VvNxohJ","trusted":true}},{"cell_type":"markdown","source":"\n \n[1] https://openai.com/blog/chatgpt The codes for submission are copied from this amazing notebook. \n \n\n","metadata":{"_uuid":"c80697c3-8f1b-417c-a852-b11363d763f4","_cell_guid":"0599e0ef-9b30-48ce-99e4-e7ea2c5de136","id":"_ZbP9S6xxohK","trusted":true}},{"cell_type":"code","source":"import pandas as pd\n\nsubmission_df = pd.read_csv(\"/kaggle/input/2023-kaggle-ai-report/sample_submission.csv\")\nsubmission_df.head()","metadata":{"_uuid":"0abbd7e1-3ab9-4fbe-bd6e-c91fd9702600","_cell_guid":"3119f6e6-0865-4fe2-9477-5e0378cc89d6","collapsed":false,"id":"xv933sCDxohK","jupyter":{"outputs_hidden":false},"execution":{"iopub.status.busy":"2023-07-06T06:13:42.115389Z","iopub.execute_input":"2023-07-06T06:13:42.116091Z","iopub.status.idle":"2023-07-06T06:13:42.180080Z","shell.execute_reply.started":"2023-07-06T06:13:42.116045Z","shell.execute_reply":"2023-07-06T06:13:42.178608Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"submission_df.loc[0]['value']='Generative AI'\nsubmission_df.loc[1]['value']='https://www.kaggle.com/timvy13/kaggle-report-2023-ai-ethics'\nsubmission_df.loc[2]['value']='https://www.kaggle.com/code/iamleonie/towards-green-ai/comments#2332328'\nsubmission_df.loc[3]['value']='https://www.kaggle.com/code/emirhanbilgic/ai-ethics-we-are-about-to-be-late/comments#2332331'\nsubmission_df.loc[4]['value']='https://www.kaggle.com/code/luckypen/actionable-ai-ethics/comments#2332335'\nsubmission_df.head()","metadata":{"_uuid":"95871a01-467c-4c4c-a878-8fd0e21de979","_cell_guid":"37e5fcb7-9dc0-4a35-8791-757c8b57a325","collapsed":false,"id":"WW6FxidUxohL","jupyter":{"outputs_hidden":false},"execution":{"iopub.status.busy":"2023-07-06T06:13:44.513148Z","iopub.execute_input":"2023-07-06T06:13:44.513554Z","iopub.status.idle":"2023-07-06T06:13:44.527649Z","shell.execute_reply.started":"2023-07-06T06:13:44.513521Z","shell.execute_reply":"2023-07-06T06:13:44.526896Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Save the submission to a CSV file\nsubmission_df.to_csv('submission.csv', index=False)","metadata":{"_uuid":"285318bd-f260-4e91-a768-d3c727625d46","_cell_guid":"28da462c-a42a-4c2e-8aa3-c0904f73e6d5","collapsed":false,"id":"tIg1SQYzxohL","jupyter":{"outputs_hidden":false},"execution":{"iopub.status.busy":"2023-07-06T06:13:48.330015Z","iopub.execute_input":"2023-07-06T06:13:48.330516Z","iopub.status.idle":"2023-07-06T06:13:48.344497Z","shell.execute_reply.started":"2023-07-06T06:13:48.330474Z","shell.execute_reply":"2023-07-06T06:13:48.342993Z"},"trusted":true},"execution_count":null,"outputs":[]}]}
\nPrivacy Issues
\n3. Challenges : AI Misconception, Generative AI Models, and Collaborative Structure for the AI Ethics
\n 3.1 Challenges with \"Anything AI packaging\" - Misconception
\n 3.2 Challenges with the Generative AI Models
\n
The encoder outputs the mean and variance of $p(Z|X)$, which will be used to compute the loss, The closer $p(Z|X)$ to a standard normal distribution $N(0, I)$ is, the smaller loss will be. The decoder plays a role of mapping the $Z$ from standard normal distribution to $p(X)$. Given a latent variable $Z$, a trained generator will output a image. 3.3 Collaborative Structure for the AI Ethics
\n
The encoder outputs the mean and variance of $p(Z|X)$, which will be used to compute the loss, The closer $p(Z|X)$ to a standard normal distribution $N(0, I)$ is, the smaller loss will be. The decoder plays a role of mapping the $Z$ from standard normal distribution to $p(X)$. Given a latent variable $Z$, a trained generator will output a image.4. Discussions : Metrics, Quality, and Lessons Learned
\n 4.1 Metric or Standard Regarding AI Ethics
\n 4.2 What Quality should the AI Ethics are actually promoting to the humanity?
\n
The encoder outputs the mean and variance of $p(Z|X)$, which will be used to compute the loss, The closer $p(Z|X)$ to a standard normal distribution $N(0, I)$ is, the smaller loss will be. The decoder plays a role of mapping the $Z$ from standard normal distribution to $p(X)$. Given a latent variable $Z$, a trained generator will output a image. 4.3 What Lessons
\n
The encoder outputs the mean and variance of $p(Z|X)$, which will be used to compute the loss, The closer $p(Z|X)$ to a standard normal distribution $N(0, I)$ is, the smaller loss will be. The decoder plays a role of mapping the $Z$ from standard normal distribution to $p(X)$. Given a latent variable $Z$, a trained generator will output a image.7. Conclusion
\n
By recognizing the need for ethical considerations, taking proactive steps, and fostering collaboration, we can shape AI technologies that are beneficial, trustworthy, and aligned with human values.
Ethics Ethics EThics
wuith AI we can create better world.Reference
\n
\n[2]\n\n","metadata":{"_uuid":"e4315b81-abf5-4b6d-8351-b70f9771e49b","_cell_guid":"c42a0149-5bc0-4be3-94d5-32a143467eab","id":"3P3fYQ4CxohJ","trusted":true}},{"cell_type":"markdown","source":"\n\n\n\n\n\nCode for submission
\n