The Many Faces of Artificial Intelligence (AI)

By Bodong Chen in musing

September 6, 2024


(Photo Credit: ChatGPT 4o)

AI is often the elephant in the room, surfacing in everything from high-stakes policymaking and academic research to casual conversations with my barber. It permeates so many aspects of life that it’s hard to escape its presence. Yet, because AI is so vast and varied in its applications, it remains a “hyper-object”—something so complex and multi-dimensional that it feels impossible to fully grasp. It impacts education, healthcare, law enforcement, entertainment, and more, each in dramatically different ways. AI is not just one thing; it’s many things, depending on where and how it is deployed.

In my field of education, AI development is moving quickly. From adaptive learning systems to automated grading tools, AI is already changing the way students learn and teachers work, for better or worse. However, these rapid developments bring both excitement and concern. How do we make sense of this fast-moving, often elusive technology?

To help us grasp—if not fully understand—AI, I’d like to take an unconventional approach: a brief lesson in the Chinese language. In Chinese, the pinyin of ai can map onto dozens of different characters, each with drastically different meanings. Similarly, as we explore the facets of AI, we can think of it as having many interpretations, some hopeful, others more cautionary. Below, I present five characters pronounced as ai, each representing a unique perspective on AI’s potential.

1. 爱 (ài): Love

AI, at its best, can be developed out of love and care. The idea of “AI for good” is about leveraging technology to nurture happiness, well-being, and progress. In education, for example, AI tools are often touted for enhancing teachers’ work efficiency, reducing administrative burdens, and preventing teacher burnout. A teacher’s day is filled with grading, lesson planning, and constant interaction with students. By automating routine tasks, AI allows educators to focus more on the relational aspects of teaching, like mentoring and personal connection, which are often lost in the grind of daily duties.

Beyond education, AI-driven healthcare tools can help diagnose diseases earlier, leading to better outcomes for patients. These are hopeful visions—AI as an aid to human flourishing. Yet, it’s crucial to design AI systems rooted in empathy, ethics, and care for the most vulnerable. Often, this work is not done by AI but by humans who bring these values into AI development.

2. 隘 (ài): Narrow

The character once described narrow, restrictive paths in ancient China, and is now more often used to mean narrow-mindedness. In the context of AI, this speaks to the limitations of current AI models. Today’s AI, particularly large language models (LLMs), follows the “connectionist” paradigm, which processes vast amounts of data to predict outcomes but lacks the depth of human cognition. AI is incredibly powerful at pattern recognition and data analysis, but it is also “narrow-minded” in significant ways.

For instance, while these models excel in tasks like content generation or voice transcription, they fall short in areas requiring intuition, context, or embodied experience. As an example, Llama 3.1, an AI language model I interacted with recently, admitted it lacks embodied cognition, contextual understanding, intuition, episodic knowing, pragmatic knowing, artistic insight, and epistemic justification. It can’t “know” the world in the way humans do—through lived experience, intuition, or sensory awareness. It favors “fast” knowledge that can be built on patterns, and is incapable of “slow” ways of knowing that takes time and involves presence and interaction. AI models privilege certain ways of knowing while sidelining others. While we marvel at AI’s capabilities (which get better each day), we must recognize its inherent limitations and be mindful of what is currently lacking.

3. 哎 (ài): Disappointment

The character 哎 often conveys disappointment, much like an exaggerated “sigh” in English. AI, in many instances, has not lived up to its grand promises. For example, AI tools in education have been marketed as transformative, but the results often fall short of expectations. Adaptive learning platforms, touted to tailor education to individual students’ needs, sometimes fail to accurately capture the nuances of how students learn. These systems may overemphasize quantitative assessments, like multiple-choice questions, while neglecting higher-order competencies like critical thinking or creativity.

This sense of disappointment extends beyond education. This gap between expectation and reality is a reminder that AI is still in its infancy—and that the hype surrounding it often exceeds its current capabilities.

4. 碍 (ài): Obstacle

While AI has the power to enhance our lives, it can also create new barriers. The character 碍 means “obstacle,” and AI systems can sometimes function as gatekeepers, restricting access rather than democratizing it. For example, AI-driven hiring systems that filter resumes based on certain keywords can inadvertently perpetuate bias by excluding candidates who don’t fit conventional molds. Similarly, AI in education can create obstacles when students have learning preferences that are not aligned with assumptions baked into AI tools.

5. 癌 (ái): Cancer

Lastly, means “cancer,” symbolizing the darker side of AI—the biases, unfairness, and opacity that can spread and grow if left unchecked. Foundational models are still not interpretable to the most part. AI systems, running as “black boxes,” operate with layers of complexity that even their creators struggle to fully understand. As these systems are increasingly integrated into everyday life, there is a risk that they will exacerbate existing social inequalities, much like an unchecked malignancy.

If AI systems are deployed in various contexts by organizations driven by profit and competition, these issues of bias and fairness can be quickly amplified. Without transparency and ethical oversight, these problems can spread like cancer, undermining trust in AI technologies and creating harm in ways that are difficult to undo.

Conclusion

AI is a multifaceted technology with the potential for both benefit and harm. By understanding its many faces—whether as a force for love, a narrow-minded thinking machine, a source of disappointment, an obstacle, or a vehicle to spread harm—we can better navigate its complexities. The future of AI depends on our ability to approach it with diverse perspectives, ensuring that we leverage its power for good while mitigating the risks it presents. Only through thoughtful design, oversight, and collaboration can we ensure AI serves society in positive, meaningful ways.

Posted on:
September 6, 2024
Length:
5 minute read, 1028 words
Categories:
musing
Tags:
AI infrastructure
See Also:
Let's complicate the idea of infrastructure in conversations about GenAI
Infrastructuring sustainable innovations in education: A quick update
Integrating generative AI in knowledge building
comments powered by Disqus