AI in Engineering – Ethical Implications

Shilpa Prabhudesai

Engineering has always been at the forefront of technological innovation and scientific advancements. Engineering has undergone a significant transformation with advancements and innovations in artificial intelligence (AI). This has transformed processes, augmented decision-making, and enabled innovations that were previously deemed impossible. From smart cities to autonomous vehicles, AI’s footprint in engineering is expansive and continuously evolving.

Contrary to traditional tools, AI systems have the ability to make autonomous decisions, learn from historical data, and adapt behaviour. These capabilities have raised questions about transparency, social impact, privacy, and the responsibility of AI systems. Hence, with AI’s rapid integration, ethical boundaries must be identified when creating AI tools or implementing AI projects. AI ethics are the moral principles that help companies with the responsible, fair development and use of AI.

This article discusses these ethical AI implications in engineering, emphasizing the importance of AI ethics, key challenges, stakeholder responsibilities, and measures to impose ethical AI in engineering.

What is AI Engineering?

AI engineering is a multidisciplinary field that combines computer science, software engineering, and data science to design, develop, and deploy AI systems. It uses techniques such as machine learning, neural networks, and deep learning to develop practical, real-world applications of AI.

AI engineers are professionals working in the field of AI engineering. They build and train algorithms that can process data, make predictions, and automate tasks, much like a human brain. This role combines expertise in programming, data science, software development, and data engineering. Refer to AI Engineer: The Skills and Qualifications Needed for more details.

Some of the characteristics of AI Engineering are:

• Focus on Practical Applications: AI engineers focus more on practical applications and bridge the gap between research and implementation.
• System Design: AI engineering builds tools and infrastructure that support AI models that are efficient and reliable.
• Collaboration: AI engineers collaborate with software engineers, data scientists, and other business stakeholders to understand business requirements and translate them into AI solutions.
• Key Skills: AI engineers use engineering and AI concepts extensively. They have very strong programming skills with knowledge of data processing and analysis, and experience in machine learning algorithms. Read: Machine Learning Models Testing Strategies.

Image recognition systems, personalized content recommendations (like on Netflix or Spotify) are some of the examples of AI engineering.

In essence, AI engineering is turning AI research into scalable and usable technology.

The Importance of Ethical AI in Engineering

AI systems have a profound impact on individuals, communities, and societies as a whole. Hence, ethical considerations in AI development are essential. Ethical AI is necessary for the following reasons:

• Prevent Harm & Promote Fairness: AI often makes autonomous decisions based on complex algorithms and vast amounts of data. Ensuring that AI operates consistently with ethical principles is essential to prevent harm, promote fairness, and uphold human rights. For example, when analyzing the population data for job opportunities globally in the manufacturing sector, it will not be fair if AI algorithms filter out the female population with a bias that they are not fit for manufacturing jobs.
• Augment Human Intelligence: According to AI proponents, this technology is meant to replace human intelligence. This means the same issues that affect human judgment can seep into the technology. One classic example that affects human judgement is favoritism. AI systems should take care that this favoritism does not affect the analysis and judgement.
• Biased & Inaccurate Data: AI projects that are built on biased or inaccurate data have not-so-good consequences, particularly for marginalized and underrepresented groups and individuals. AI systems should be able to detect inaccuracy and bias in the data that it is going to train models on. If the data is biased or inaccurate, the results are going to be wrong. For example, if the data that is collected for implementing an automated traffic system in a city deliberately highlights only wealthy, affluent surroundings and ignores less privileged areas, the autonomous traffic system is surely going to collapse one day.
• Hasty Development: AI algorithms or models can become unmanageable if built too hastily, and correcting learned biases may be challenging.
• Integration of AI Technologies: This integration in development, deployment, and societal environments has a profound influence and, hence, ethical implications. For example, it is ethical to integrate AI systems to promote religious bias or affect communal harmony. So, when integrating AI technologies, due attention should be given to the aspect that this integration will not harm society.
• Set Ethical Boundaries: Ethical implications ensure that AI systems operate within ethical boundaries and reflect the values and interests of their stakeholders. For example, if you provide a prompt related to instigating communal tension in a particular city to ChatGPT, it should promptly reject the request and not provide answers to these prompts just because it’s the job. It is always better to set ethical boundaries when working.
• Responsible AI: Ethical implications bring in public trust in AI systems across domains and mitigate potential risks and liabilities, thus promoting responsible use of AI.

AI in Engineering: Challenges

AI-driven engineering faces key ethical challenges that may affect the project adversely. These challenges should be addressed appropriately. Let us discuss these challenges here and also how to tackle them:

1. Bias and Discrimination

AI systems are driven by data and algorithms, and hence, they are good if the data is good. If the data collected is inaccurate and does not represent the population, decisions taken using this data may be biased. The outcomes, in this case, are discriminatory. For example, an intelligent traffic control system may collect data only to traverse affluent neighborhoods, leading to increased congestion in specific areas. Another example of discrimination is when the data for HR systems is filtered with the word “Women”. In this case, the outcome will be a clear discrimination against women. To prevent these biases and discrimination, ethical engineering must ensure that fairness and inclusivity are promoted in AI systems.

2. Accountability and Liability

Determining responsibility when things go wrong is a pivotal ethical dilemma in AI. For example, if an AI-powered diagnostic tool makes a mistake in diagnosis and the patient is harmed, who should shoulder the blame? Is it the tool that performed the diagnosis, the technician who used it, the company that manufactured it, or the engineers who conceived and deployed it? The correct stakeholders must be held accountable. Thus, the AI development process should establish a clear line of accountability and transparency in the system so that stakeholders know who bears responsibility for potential consequences and how the decisions are made. For this purpose, decision-making processes should be documented, and traceability of AI behavior must be ensured. Also, safeguards that allow human oversight must be implemented.

3. Transparency and Explainability

AI systems are often opaque and complex, operating as “black boxes”, making it difficult to understand how they make decisions without offering insights into their reasoning. This lack of transparency poses an ethical challenge, especially in high-stakes industries. For example, in aerospace industries, the manufacturing of warplanes may not be transparent enough and may lack explanation. Hence, if some other team takes over the manufacturing in the future, they will not know the details of the warplane. Also, since the system is not explainable enough, end-users may find it difficult to use the warplanes. AI systems should be explainable, and engineers, regulators, and end users should understand their decision-making process. When a system is explainable and transparent, it builds trust, and stakeholders can make more informed decisions when needed. Read: What is Explainable AI (XAI)?

4. Privacy and Surveillance

AI collects, analyzes, and interprets information from large volumes of data, which can interfere with users’ privacy and security. Engineering applications such as surveillance systems, Internet of Things (IoT) devices, and smart cities are prone to privacy violations. For example, in the case of smart cities, an AI system may collect hospital data. While collecting hospital data, many patients may be violated by exposing their sensitive data. AI systems must comply with data protection laws and apply principles of privacy-by-design. Ethical design principles include minimal data collection, secured data storage, and user control over personal information. In addition, strong data protection measures such as encryption and anonymization should be incorporated into the AI system. Read: AI Compliance for Software.

5. Job Displacement and Economic Impact

AI enhances productivity and reduces costs through automation. However, it may also lead to job losses and economic impact. It may displace the roles that were previously considered sage in the engineering sector. For example, in recent years, advanced chatbots have been used to screen candidates for job opportunities, so there is no need for HR representatives to make a call and screen candidates. Ethical AI should consider the social consequences of AI systems and work towards mitigating the adverse effects through techniques such as retraining, redesign, and inclusive development. Read: Chatbot Testing Using AI – How To Guide.

6. Security and Misuse

Like any other technology misused, AI can be used for malicious purposes such as deepfakes, cyberattacks, and surveillance. To minimize this, frameworks should have robust security measures and ethical principles. For example, autonomous weapon systems may raise existential and humanitarian concerns. AI should handle the data in a way that not only protects the data from breaches but also ensures that users have control over the usage of data, so that it is not used for malicious purposes. Read: Top 10 OWASP for LLMs: How to Test?

7. Human Safety

AI systems must prioritize human safety and consider all the principles and measures to ensure the system is designed and developed accordingly. Relying on AI for decisions in healthcare, hiring, and criminal justice can reduce individuals to data points and also risk their safety.

For example, an oncology pathology platform that uses AI to diagnose cancer disease may harm the patient if it fails to diagnose the exact stage of the disease or the organs that are affected by the disease. AI systems, especially the medical systems may thus jeopardize human life. People often are not aware that their personal information is used for other purposes especially in automated decision-making systems.

8. Existential Risks

Though still not a concrete idea, AI may threaten human existence, as it is speculated that AI may become self-aware and surpass human intelligence. There are also many experiments going on currently on ideas to have human emotions in AI models. Right now, AI models and robots perform as per the instructions given to them by programs. If they start thinking and have emotions of their own, there will be a day when AI robots will replace or harm humans. Thus, humans also face an existential challenge from AI.

Role of Stakeholders in AI Ethics

Designing and deploying AI ethical principles is not the company’s responsibility alone. It requires close collaboration between industry personnel, business leaders, and government representatives. All stakeholders must examine how AI interacts with social, economic, and political issues and determine how machines and humans can exist harmoniously, reducing potential risks and ensuring human safety. Each of the stakeholders has the responsibility to ensure less bias and risk for AI technologies and create a more ethical AI:

• Engineers and Developers: They uphold safety, welfare, and public rights. So, engineers must ensure the AI system is robust, transparent, and fair. They should always be well informed about the ethical implications of AI and integrate them throughout the design and development process.
  Principles emphasized by professional organizations such as IEEE and NSPE, for ethical conduct like the imperative to avoid harm and promote societal good, should be the guiding principles for AI system design and development.
• Academicians: People in the education and academic field, such as researchers and professors, are responsible for developing theory-based statistics, ideas, and research that can support governments, corporations, and non-profit organizations. Academic institutions are also responsible for preparing the next generation of engineers to navigate AI’s ethical complexities. The educational curriculum should include ethical training, case studies, etc.
• Corporate and Employers: Big corporations and employers are responsible for creating ethics teams and an AI code of conduct. They should set a standard for the companies to follow. Companies should bring in ethical AI development by taking adequate measures such as conducting fairness audits, promoting transparency, and ensuring that AI systems align with corporate social responsibility goals.
• Government: Legal and ethical frameworks for AI should be set up in collaboration with government agencies and committees. Rules and regulations must balance innovation with people’s safety and rights. Governments should set standards for data use, mandate impact assessments, and support ethical AI research. It is also necessary to address global challenges, especially with respect to the arms race and international surveillance.
• Intergovernmental Entities: Entities such as the United Nations or the World Bank play an important role in measures such as raising awareness and drafting universal agreements on AI ethics.

Way Towards Ethical AI in Engineering

To truly address AI’s ethical implications, the engineering community must cultivate a culture of ethical awareness and proactive engagement. This requires:

• Ethics by Design: Every stage of the engineering lifecycle, from problem definition and data collection to deployment and maintenance, should have ethical considerations integrated into it. Development and deployment of AI systems should be governed by ethical frameworks and guidelines.
• Diverse Datasets: Try to employ diverse data sources and techniques, such as data augmentation, to create synthetic data to fill the gaps in existing datasets.
• Bias Mitigation Techniques: Use bias detection and mitigation techniques to detect and correct biases in AI algorithms.
• Interdisciplinary Collaboration: Engineering and other departments, such as legal, HR, etc., should be involved in shaping AI systems to ensure they are aligned with human values and needs.
• Ongoing Monitoring and Evaluation: Engineering teams should implement feedback loops, impact assessments, and revise systems wherever necessary. They should also continuously monitor AI systems for unplanned consequences and ensure continued alignment with ethical principles.
• Empowerment and Advocacy: Engineers should have access to ethical training, legal support, and advocacy channels, and be empowered to speak out against unethical practices. Ethical oversight and auditing processes should also be established to ensure the moral nature of AI systems.
• Human-Centric Design: Always prioritize human-centric design to ensure AI systems are designed with human values and requirements in mind. Engineers should address these concerns proactively and contribute to the development and deployment of AI systems, making them responsible, beneficial, and socially aligned.

Case Studies Demonstrating Ethical Issues

1. ChatGPT

A classic example that demonstrates ethical issues is ChatGPT. This AI tool helps users create original content by asking questions or providing prompts. The model is trained on data from the internet, and it can answer questions in various ways, whether it is an article, poem, essay, code, or proposal.

However, people use ChatGPT to win essay competitions or coding contests. This is a violation of ethical standards and may indirectly harm genuine participants. In such cases, ethical AI principles should be established to prevent such a dilemma.

2. Smart Infrastructure and Equity

Smart city projects use AI to optimize traffic, public services, educational institutions, and energy. However, it is important to incorporate diverse perspectives when doing this, or it can marginalize communities. For example, public services may be diverted to affluent neighborhoods only. Engineers in such cases must advocate for equitable AI deployment and seek public consultation in the design process.

Conclusion

With the rise of AI, the engineering sector has witnessed tremendous growth and opportunities, as well as ethical challenges. Engineers must ensure fairness and transparency, protect privacy, and minimize harm. They also play a crucial role in shaping the future of AI. Ethical engineering upholds values of responsibility, integrity, and human dignity in addition to compliance. As AI develops, engineers must rise to the occasion, not only as innovators but as stewards of technology committed to the public good.

Integrating ethical principles into the core of engineering practice can protect the rights and well-being of individuals and society at large.

“We spent so much time on maintenance when using Selenium, and we spend nearly zero time with maintenance using testRigor.”

Keith Powe VP Of Engineering - IDT