Ethical Governance of AI
Ethical governance of artificial intelligence is the framework, policies, and regulations that ensure the application or use of AI with safety, responsibility, and fairness without harm to the nature of human beings and animals. AI, however, is beneficial for all areas of human interest; it can cause harm to nature as well as human lifestyle. It can be utilized in the worst situations for people whose purpose is against human well-being. It is essential to implement ethical governance of AI in such a manner that the AI reflects the benefits and safety only.
Example
1. AI in the selection of a candidate or a shortlist of candidates must explain why someone is rejected.
2. AI refers to the advancement in the machine that enables it to think like a human and behave like a human being. While AI can achieve greater accuracy and productivity than humans in certain areas, its increasing adoption of human-like characteristics may lead to potential negative impacts.
3. AI in content generation avoids providing information for making a bomb.
Principles of Ethical Governance of AI
Transparency
AI – Artificial Intelligence must be explainable and provide alternatives for any solutions. The final choice must be made by a human being to avoid unnecessary results.
Example
1. AI in banking, specifically for loan approval tasks, may reject the application with an explanation of the reason why the loan applicant is rejected.
2. Content-creating platforms require users to provide only the necessary information and to avoid including any unnecessary details.
Fairness and Non-Discrimination
As AI does not belong to a gender, culture, religion, or society, it cannot implement discrimination based on gender, religion, etc. It provides a fair evaluation and judgement of human issues.
Example
AI does not prefer any one gender for the job, nor does it shortlist males or females when an organization uses AI for the recruitment process. Irrespective of gender, culture, or religion, the AI selects the candidates based strictly and accurately on criteria set by the organisation. When a human being performs the same work, discrimination may occur.
Accountability
AI, however, belongs to machines; it is made by humans. The human or organization must be responsible for any impact or influence, whether positive or negative, caused by AI. A human or an organization that belongs to the AI must be answerable for any consequences.
Example
The recruitment process, when executed by AI, may select the candidate who may not be suitable for the requirement due to the criteria set by AI. The organization or human must be answerable for the result of recruitment. Blaming the AI is not fair and ethical.
Privacy and Data Protection
Privacy and data protection are compulsory and necessary to ensure that data collection and storage are done safely and used for the intended purposes only.
Data is a significant input for power in today’s world. In every field, data plays a vital role in achieving dominance. In business, politics, economics, defence, research, and the like, many areas depend more on data. So the privacy of the data is an unavoidable mandatory factor for all these areas. AI needs data to run with excellence. If data privacy is not protected by AI, then it might be possible to make confidential data available for the people who use it positively and for the people who can use it negatively. Negative use of data can lead to the worst consequences.
Example
1. ChatGPT does not provide data on how to make a chemical bomb. Platforms like ChatGPT, Gemini, and DeepSeek provide large amounts of data tailored to user requirements. These platforms provide data that precisely matches the user’s requests.
These content creation platforms are completely implementing privacy and data protection. All users protect and maintain the confidentiality of their data. These platforms provide data that is not susceptible to negative use.
2. Consent-based data collection
Many AI apps, when a user uses them, may ask for consent to allow location access and other access.
3. Data minimisation: AI collects the necessary data, not extra data collected from users. For example, cab apps like OLA, TADA, Gojek, and ComfortDelGro cap service providers that collect current location data to reach the location. It does not collect contacts, photos, and other details from the phone.
4. End-to-End Encryption in AI Chat Systems: This feature is mostly common, and most people are aware. Actually, the Meta company promotes this feature to make people aware of it and use this app more safely and freely.
Safety and Security
The AI system does not behave wrongly; however, there are some errors or failures. It avoids any malfunction or incorrect, harmful behaviour that causes negative results or outcomes.
Example
AI in medical diagnosis tools is beneficial, but failures or errors are harmful. Actually, such AI is error-free and secured from failure, hacking, and any other harmful behaviour.
Human Oversight (Human-in-the-Loop)
Regardless of how well AI performs a task, the final decision must be made by a human. AI should act as support, not replace humans. This feature is very necessary to protect human importance and any unexpected results of artificial intelligence.
Example
1. While making a credit score, the AI plays a vital and excellent role and determines the credit score of the individual accurately. However, AI only performs this task; a human officer must make the final decision.
2. While the AI accurately analyses the patient, doctors must make the final diagnosis.
Beneficence (Social welfare)
AI, artificial intelligence, must be utilized and made for social benefits. It must contribute to human well-being. The ultimate goal of any artificial intelligence must be social welfare or human well-being, either directly or indirectly.
Example
Artificial intelligence, or AI, in the fields of agriculture and weather provides accurate data about rainfall and helps farmers make the right decisions about crops and practices.
Inclusiveness
AI – Artificial Intelligence must be easily accessible to all people, including rural and low-income populations. The purpose of any technology has always been human well-being and improvement in all aspects of human life.
Example
ChatGPT, Gemini, DeepSeek, and the like, and many other content creation AI platforms are available for all. Actually, these apps are becoming the alternative to Google. The key difference between Google and these platforms is the variety of data they provide. Google provides the data from a number of blogging sites. On the other hand, these apps only offer a limited amount of essential data. Actually, the data provided by the AI apps is from all the online sources. It summarises it and provides it as per user requirements or demands. However, Google provides 100 links that are related to user requirements.