Explorations in Ethical AI

Within my master's program, I have the privilege of delving into thesis topics that truly ignite my passion. My chosen focus revolves around the profound influence of AI and machine learning on the dynamic realms of design and digital media. Furthermore, I am keen to explore how these technologies shape the current landscape and the potential transformations they hold for the future. Equally intriguing is their impact on the daily lives of individuals and families.

My objective is to center my attention on the multitude of advantages that AI has brought about over the past decade. I intend to gain a comprehensive understanding by synthesizing data and observations, thereby acquiring valuable insights. Additionally, I aim to derive perspectives from leading brands and everyday products that harness and innovate with AI to better our world.

In this era marked by swift advancements in AI technology, we embark on a journey into the intricate domain of AI ethics. We delve into the convergence of innovation and responsibility, exploring aspects ranging from decision algorithms to societal implications. During the recent semester, my team and I delved deep into the examination of the point where AI's capabilities align with ethical principles, with a particular emphasis on the fundamental foundations of trust, transparency, and fairness.

Meet AIDEN

AIDEN, a cultural probe identifier designed to collect data from individuals across various segments, including workplaces, social circles, family, as well as our social and academic networks. We employed AIDEN to pose inquiries via a predefined smart form, aiming to gain an extensive understanding of people's perceptions of AI in the contemporary context. We garnered an impressive response, receiving over 100 participant inputs, which provided a diverse range of insights spanning multiple generational divides.

Over 45%

Over 45% of those polled and interviewed are Reluctantly Adapting to AI and its prevalence in our society. In general people don’t trust it because they don’t know how it works and therefore don’t know how safe it is.

Due to the extensive nature of our observations and findings, we possessed an abundant volume of data and perspectives to focus on. These perspectives include an optimistic viewpoint, a cautious standpoint, and a more informative and forward-looking perspective. We encouraged our audience to maintain an open mindset as AI significantly impacts our daily lives, both consciously and subconsciously. Just as the industrial revolution brought about transformative changes in our culture, AI has been undertaking a similar role for the past six decades or more.

Our Process

Our methodology commenced with the construction of a stakeholder map encompassing experts, developers, and users. Subsequently, we assimilated all the insights garnered from these interviews to facilitate the process of Affinitization.

This technique involves amalgamating the viewpoints, perspectives, and insights of a group of individuals well-versed in the subject matter. This comprehensive data collection phase served as the foundation for the development of our Affinity diagram.

Following this, we conducted numerous one-on-one interviews with our users, developers, and experts, involving more than 20 questionnaires. We also launched a cultural probe in the form of a social media persona quiz. The culmination of these efforts resulted in the identification of our primary data points.

From there we synthesized our key themes and insights such as:

• The current importance of AI and its continued progress highlight the necessity for careful planning, comprehension, and investigation.

• Starting an AI transition for fresh concepts, simplified procedures, and improved accessibility.

• Unbiased AI tools made with a focus on adaptability and good results are produced by mindful innovation.

• AI enables novel environments and ideas that enhance humanity by extending human engagement across distances.

• Addressing ethical oversight and change to address AI dangers like misinformation, regulatory gaps, privacy concerns, and prejudices.

The Framework

The Significance of Security for Genuine Enjoyment and Well-being

Emphasizing safety should consistently take precedence when integrating AI technology. Placing safety at the forefront enables AI systems to avert accidents and reduce potential risks. This fosters confidence in the technology and elevates the overall user experience by establishing a safe and dependable environment.

61%

Embrace the potential benefits of AI

Simplify the Complex

78%

Prioritize Privacy

93.75%

Considered themselves to be a “Basic User” despite their industry title or experience.

Inform the User

Problem

Users lack insight into the utilization of their data, its handlers or corporate recipients, the purposes it serves, and how it can contribute to enhancing AI-driven outcomes as per their preferences.

Solution

Enhanced transparency regarding data sharing, potential uses, purchase possibilities, and easy cancellation options at any moment. Additionally, empowering the public to have the capability to delete their data.

Protect Everyone’s Data

Problem

Users are aware that various digital platforms can potentially expose their personal information, making them susceptible to scams or security vulnerabilities.

Solution

Provide an explanation of data protection measures and the legal regulations in place to enforce this protection while also ensuring companies are held accountable.

Include More Edge Cases

Problem

Users are often excluded from discussions concerning the utilization of their data, such as in the context of artists and generative AI.

Solution

Establish and uphold ethical regulations that safeguard the interests of artists and enable the public to enjoy using AI as a tool when required.

To ensure a clear and concise grasp of our interviewees and the gathered data,

we posed the following question:

How might we improve how people experience artificial intelligence across the world?

Insight

A key insight we gained centered around privacy and security emerging as the foremost concern. This prompted us to analyze feedback holistically, understanding how feelings of safety, privacy, and security significantly shape AI's evolution and its cultural impact, while also driving ongoing innovation.

Solution

Our initial approach involves establishing an autonomous private enterprise to tackle these concerns and enhance public safety. We've named it the "Equitable Technology Agency," or ETA for short. This independent entity will be composed of individuals from diverse demographics and AI expertise, fostering an inclusive, ethical, and cooperative influence on AI's present and future development. Our aspiration is that this company eventually forms a unified and foundational framework for a forthcoming sector, one with the ability and expertise to extend safety and knowledge on a broader scale.

“Personally, privacy and security are the main concerns based on the initiatives I would use it for. I feel there is no structured regulation with AI at this point and there are vulnerabilities. A bias issue also being one of them.”

- Genya Adesso Vice President, Marketing at Bally Sports Florida | Sun Regional Sports Network

Tone

The ETA not only embodies its mission of integrity through trust and transparency, but it also holds great significance from a branding and personality standpoint. Its public communication style is designed to foster trust and safety, seamlessly combining authenticity, compassion, expertise, and transparency. This approach is fundamental to the heart of the business mission and its commitment to the public.

How does it align with data

ETA's foundation lies in data and portrays trust through its branding. In this context, you can observe the harmonization of where and how we integrate the concepts of trust, empathy, and ethics into its branding. This is achieved by communicating these values to users through its visual representation and user experience. To sum up, the core of ETA's mission is to contribute to a safer AI-driven world while establishing trust through safety, transparency, ethics, and optimization.

Industries We’ve improved

At the heart of ETA's mission to enhance AI's impact on the world by ensuring safety and fostering trust, here are several concrete outcomes showcasing how AI is influenced and, most importantly, enhanced.

Environmental  & Sustainability Sectors

The Environmental & Sustainability Sectors, particularly in the context of endangered species, have showcased the positive outcomes achieved through the utilization of AI. A case in point is the work of neural artist Sofia Crespo. She employs AI to craft artistic renditions of chimeras, blending actual images of endangered species from our world to generate entirely new creations. Her purpose is to inspire concrete conservation actions.

In an exploratory venture, Crespo brings forth creatures solely born from AI, addressing the challenge of insufficient data and representation for lesser-known endangered species, such as the boreal felt lichen, which often lack visibility. By concentrating on these overlooked endangered species, she strives to bring attention to their existence. It's vital to acknowledge that the accuracy of her AI-generated images hinges entirely on the available data.

In her pursuit, she commits herself to using AI as a tool by collecting comprehensive data, identifying patterns, and constructing impartial datasets. The ultimate aim is to offer fair and accurate depictions, thereby educating the global community about endangered species and creating a tangible, positive impact through AI's application.

This insight points toward a more optimistic future concerning crucial issues like deforestation, climate change, species extinction due to human intrusion, and farming practices. The prospect emerges that these challenges, which jeopardize our planet's well-being, could be alleviated for the benefit of forthcoming generations through this scientific approach. The emotional resonance extends beyond mere hope; it holds a profound impact that reaches far into the future.

Global Safety

Creating a Global GDPR (General Data Protection Regulation) framework involves implementing extensive privacy laws that span sectors and cover businesses of all sizes. This framework supersedes the Data Protection Directive 1995/46 and has a main goal: to institute regulations that protect personal data and data mobility. This prioritizes fairness, trust, and an impartial approach, upholding AI's ethical purpose for the greater good and curbing bias through regulatory measures and safety protocols.

Within these regulations, an AI transparency law would be enacted, mandating all AI systems and platforms to provide clear information about their abilities, limitations, and underlying algorithms. Users should be informed when engaging with an AI system, and developers must openly share data details for model training and potential biases. For example, an AI Privacy Law would ensure AI systems follow strict privacy principles, securely handling user data only for its intended uses. Unauthorized storage of personal data would require explicit consent, and AI deployment should respect personal privacy and refrain from malicious data manipulation.

The framework would also include an AI Accountability Law, making developers and organizations responsible for their AI systems' consequences. Mechanisms for investigation and rectification would be established in case an AI system causes harm or significant errors. An AI Bias Mitigation Law would oblige developers to proactively identify and address biases in their systems. This involves diverse and representative training data and continuous monitoring to prevent harmful biases from persisting.

Furthermore, an AI Morality and Ethical Compliance Law would stipulate that AI systems must be ethically designed and adhere to universally accepted moral principles. Activities that could jeopardize human safety or well-being, involving harm, deception, or malicious intent, would be strictly prohibited.

These laws are designed to create a framework that promotes responsible AI development and usage while shielding humanity from potential risks. As AI technology advances, these regulations prioritize human welfare and prevent the misuse of this potent technology. This realization sparks an emotional connection of trust and hope, stemming from the positive potential implications for the future.

Education  & Training

Education, a precious asset, is unfortunately not equally accessible to all. Uneven learning abilities and unequal access to education are realities. In the domain of artificial intelligence (AI), education holds significant importance. The focus centers on how educational systems and educators can synergize with AI in a manner that harnesses its potential to enhance learning, not hinder it.

AI stands poised as a source of inspiration, demonstrating how it can assist students requiring extra support on their learning journey. By involving students in pre-classroom tests, questions, and narratives, AI has the potential to greatly enrich education and promote inclusivity.

The introduction of a personalized AI-driven educational approach yields numerous benefits. It empowers students to engage in practical problem-solving experiences utilizing algorithms and controlled permissions designed to stimulate critical thinking. Simultaneously, educators gain fresh tools to craft and implement innovative lesson plans. This approach encourages the fusion of diverse materials and fosters the exploration of novel educational systems tailored to individual interests and criteria.

Embracing this tailored educational model paves the way for students and educators to embark on a transformative voyage of enriched learning and intellectual growth. In the long run, harnessing AI as an educational tool can expand accessibility to encompass all, transcending select individuals and extending equal opportunities to everyone.

Presently, Sal Khan, Khan Academy's founder and CEO, envisions artificial intelligence as a catalyst for the most remarkable positive change in education. He highlights the prospects for students and educators to collaborate with AI tools, envisioning personal AI tutors for every student and AI teaching assistants for every educator. He also demonstrates exciting new features of their educational chatbot and AI-based learning through Khan Academy's innovative approach.

Medical

Much like the various impact levels discussed earlier, another insightful observation applies to the realm of medicine. In healthcare, AI stands to significantly lower societal healthcare costs by minimizing overhead expenses and delivering precise diagnoses. Moreover, it has the potential to democratize medical access, extending its advantages to those without the means or health insurance coverage.

A relevant illustration of AI's role can be seen in dermatology. As an augmentative tool, AI supplements the expertise of highly trained specialists like dermatologists, who command substantial annual salaries of about $400,000. The path to becoming a skilled dermatologist requires over a decade of training. Integrating this expert knowledge with machine learning enables the early detection of melanoma, one of the deadliest forms of skin cancer. This is particularly crucial for individuals without easy access to dermatologists or healthcare services.

Stanford University's work underscores this potential. Training AI on a dataset of 129,000 skin condition images, including melanoma and carcinomas, allows AI to perform comparably to top-tier human dermatologists. This was confirmed through a separate dataset presented to 25 board-certified dermatologists of Stanford's caliber. In most cases, AI either matched or exceeded the classification accuracy achieved by human dermatologists.

In the medical field, AI could aid in treating presently incurable diseases, offer more accurate outcomes, and predict early indicators of future illnesses, enabling proactive prevention strategies. The future holds promise for leveraging AI to advance prevention, diagnosis, and staging, all while ensuring accessibility through tools and programs, especially for those with limited access to medical resources.

Feedback Loop

Establishing a more streamlined system of feedback loops for optimizing AI generation aims to create accessible channels for users to share their experiences and provide concrete feedback, which can then be utilized for ongoing improvements. This approach empowers users of the AI program by affording them a sense of ownership and influence in enhancing the generative AI processes. The outcomes of this endeavor would undergo scrutiny by engineers, technological experts, professionals, and developers to drive refinement.

The consensus among various groups is that AI has not yet reached a stage where entrusting it with human lives is viable. Such agreement on a matter is rare, highlighting the extent of the challenge. In this context, AI serves as a bridge between human interaction and developer involvement. Beyond this bridging role, AI's contribution is instrumental in cultivating trust and fostering a sense of ownership.

The ultimate goal is to create an environment where AI development benefits from the insights of users and experts, steadily advancing while engendering a growing rapport between humans and the technology.

AI has the potential to drive positive changes and enhance accessibility in various aspects of life, including work and personal life. However, many people are hesitant due to its novelty, which can lead to fear rather than embracing its potential for greater equality and better outcomes for all. AI's transformative capabilities depend on factors like relevance, connection, advocacy, and longevity. It is crucial to ethically develop AI during this era of continuous expansion, maximizing its potential while safeguarding humanity's core values. Together, we embark on a timeless journey, fostering bonds and inclusivity to create a lasting legacy of development and influence.

In our research, we've found that the majority of people are open to the dynamic nature of AI as long as they understand the level of control and influence it has on their daily lives. Some fear AI due to a lack of knowledge and its rapid evolution. Nevertheless, AI has become an integral part of our daily lives, from smartphones to voice assistants like Alexa, and even in medical treatments. It's a tool that continues to shape our lives, and with it, we will continue to evolve and use it to our advantage.