AI in Society
The blog is connected to my role of the head of the National Institute on AI in Society
at California State University Sacramento. However, opinions and positions expressed therein are mine, and do not represent the university's opinions or positions.
Friday, August 23, 2024
Filling Voids, Not Replacing Human Experts
Consider healthcare, particularly in underserved areas. It is a truism that a qualified doctor’s advice is better than anything an AI could provide. But what if you live in a rural village where the nearest doctor is hundreds of miles away? Or in a developing country where medical professionals are stretched thin? Suddenly, the prospect of AI-driven medical advice does not seem like a compromise; it feels like a lifeline. While AI lacks the nuanced judgment of an experienced physician, it can provide basic diagnostics, suggest treatments, or alert patients to symptoms that warrant urgent attention. In such scenarios, AI does not replace a doctor—it replaces the silence of inaccessibility with something, however imperfect.
Another case in point is mental health counseling. In many parts of the world, even in affluent countries, mental health services are woefully inadequate. Students at universities often face wait times ranging from weeks to months just to speak with a counselor. During that limbo, the option to interact with an AI, even one with obvious limitations, can be a critical stopgap. It is not about AI outperforming a trained therapist but offering a form of support when no other is available. It can provide coping strategies, lend a sympathetic ear, or guide someone to emergency services. Here, AI does not replace therapy; it provides something valuable in the absence of timely human support.
Education offers another case for AI’s gap-filling potential. Tutoring is an essential resource, but access to quality tutors is often limited, mainly because it is expensive. Universities might offer tutoring services, but they are frequently understaffed or employ peer tutors. Office hours with professors or teaching assistants can be similarly constrained. AI can step into this void. Chatting with an AI about a difficult concept or problem set might not equal the depth of understanding gained from a one-on-one session with a human tutor, but it is unquestionably better than struggling alone. AI does not compete with tutors; it extends their reach into spaces they cannot physically or temporally cover.
The same logic applies to a range of other fields. Legal advice, financial planning, career coaching—all are areas where AI has the potential to add significant value, not by outstripping human expertise but by offering something in environments where professional advice is out of reach. Imagine a low-income individual navigating legal complexities without the means to hire an attorney. An AI could provide at least basic guidance, clarify legal jargon, and suggest possible actions. All of it must be done with proper disclaimers. It is not a substitute for legal representation, but it is a world better than the alternative: no help at all.
In embracing this non-competing stance, we shift the narrative. The role of AI is not to replace human experts but to step in where human services are scarce or nonexistent. The true potential of AI lies in its ability to democratize access to essential services that many people currently go without. When AI is viewed as a bridge rather than a rival, its utility becomes much more evident. AI does not have to be better than a person to be valuable; it just should be better than the void it fills.
Monday, August 19, 2024
The Right to Leapfrog: Redefining Educational Equity in the Age of AI
AI’s potential in education is clear, particularly in how it can assist students who struggle with traditional learning methods. It is broadly accepted that AI can help bridge gaps in cognitive skills, whether due to dyslexia, ADHD, or other neurodiverse conditions. Yet, the utility of AI should not be confined to specific diagnoses. Insights from decades of implementing the Response to Intervention (RTI) framework reveal that regardless of the underlying cause—be it neurodiversity, trauma, or socioeconomic factors—the type of support needed by struggling students remains remarkably consistent. If AI can aid students with reading difficulties, why not extend its benefits to others facing different but equally challenging obstacles? Equity demands that AI’s advantages be made accessible to all who need them, regardless of the origin of their challenges.
This brings us to a deeper issue: the rigid and often unjust link between procedural and conceptual knowledge. Traditionally, lower-level skills like spelling, grammar, and arithmetic have been treated as prerequisites for advancing to higher-order thinking. The prevailing notion is that one must first master these basics before moving on to creativity, critical thinking, or original thought. However, this linear progression is more a product of tradition than necessity. AI now offers us the chance to reconsider this approach. Students should have the right to leapfrog over certain lower-level skills directly into higher-order cognitive functions, bypassing unnecessary barriers.
Predictably, this notion encounters resistance. Rooted in the Protestant work ethic is the belief that one must toil through the basics before earning the right to engage in more sophisticated intellectual activities. This ethic, which equates hard work on mundane tasks with moral worth, is deeply ingrained in our educational systems. However, in an age where AI can handle many of these lower-level tasks, this mindset seems increasingly obsolete. Insisting that all students must follow the same sequence of skills before advancing to higher-order thinking is not just misguided; it is a relic of a bygone era. If AI enables students to engage meaningfully with complex ideas and creative thinking from the start, we should embrace that opportunity rather than constrain it with outdated dogma.
The implications of this shift are significant. If we recognize the right to leapfrog over certain skills, we must also acknowledge that traditional educational hierarchies need to be re-examined. Skills like spelling and grammar, while valuable, should no longer be gatekeepers for students who excel in critical thinking and creativity but struggle with procedural details. AI offers a way to reimagine educational equity, allowing students to focus on their strengths rather than being held back by their weaknesses. Rather than forcing everyone to climb the same cognitive ladder, we can enable each student to leap to the level that aligns with their abilities, creating a more personalized and equitable educational experience.
This rethinking of educational equity challenges deeply rooted assumptions. The belief that hard work on the basics is necessary for higher-level achievement is pervasive, but it is not supported by evidence. In reality, cognitive development is driven more by engagement with complex ideas than by rote mastery of procedural skills. AI provides the tools to focus on these higher-order skills earlier in the education, without the traditional prerequisite of mastering lower-order tasks.
Moreover, the concept of “deskilling” is not new. Throughout history, humanity has continually adapted to technological advances, acquiring new skills while allowing others to fade into obscurity. Today, few people can track animals or make shoes from anymal skin—skills that were once essential for survival. Even the ability to harness a horse, once a common necessity, is now a rare skill. While some may lament these losses, they are also a reminder that as society evolves, so too must our educational priorities. Just as technological advancements have rendered certain skills obsolete, AI is reshaping the skills that are most relevant today.
As we move forward, educators must rethink how learning experiences are designed. Rather than viewing AI as merely a tool for accommodating deficits, we should see it as a means of expanding possibilities for all students. By enabling learners to bypass certain skills that are no longer essential in an AI-driven world, we can better align education with the demands of the 21st century. This is about acknowledging that the path to learning does not have to be the same for everyone. In a world where AI can democratize access to higher-level cognitive tasks, the right to leapfrog is not just a possibility—it is a necessity for equitable education.
Friday, August 9, 2024
Authorship, Automation, and Answerability
In the ongoing debate about the ethical use of AI, two main concerns stand out—one superficial and one profound. The first concern, often highlighted, is about the authenticity of authorship, with fears that AI-generated content might mislead us about who the true author is. However, this worry is largely misguided. It stems from a historically limited, Western-centric notion of authorship that blurs the line between the origin of ideas and the craft of their representation.
Take the legacy of Steve Jobs. He wasn’t celebrated for personally assembling each iPhone, but for his vision and design that brought the device to life. In our industrial world, the act of making things is not inherently authorial—designing them is. Why should it be any different with text, code, or images? If I designed this text, and used advanced tools to produce it, why am I not still the author? The shock many feel towards AI’s ability to generate content is akin to the upheaval experienced by 19th-century bootmakers during the Industrial Revolution. Automation has simply extended its reach into the realms of writing, coding, and art. The craftsmanship is replaced by automation, but the core principle remains: take pride in the ideas, not in the mechanics of their production. There is no inherent authorship in the latter.
But here’s where Mikhail Bakhtin’s notion of answerability helps our understanding of the true ethical stakes. While responsibility is often about fulfilling obligations or being held accountable after the fact, answerability is about our ongoing, active engagement with the world and the people in it. It is not just about who gets credit for the content; it is about recognizing that every action, every word, and every piece of AI-generated content occurs within a web of relationships. We are answerable to others because our creations—whether authored by human hands or machine algorithms—affect them.
The real concern, then, lies in the issue of answerability. AI-generated content often appears polished, convincing, and ready for immediate consumption. This creates a dangerous temptation to release such content into the world without thorough scrutiny. Here is where the ethical stakes rise significantly. AI may produce work that looks and sounds credible, but this does not guarantee that it is unbiased, meaningful, or truthful. It maybe garbage polluting the infosphere at best, or an outward harmful fake at worst. The ease of content creation does not absolve us of the responsibility to ensure its quality and integrity, and more importantly, it doesn’t free us from the answerability we have to the world around us.
This is the message we need to instill in our students, professionals, and anyone working with AI: you are still accountable and answerable for what you produce, even if a machine does the heavy lifting. Releasing AI-generated content without critical evaluation is akin to conjuring a spell without understanding its consequences. Like a magician wielding powerful but unpredictable magic, or a novice driver behind the wheel of a truck instead of a bicycle, the stakes have been raised. The tools at our disposal are more potent than ever, and with that power comes a heightened level of answerability.
In essence, the ethical debate surrounding AI shuold not be about the authorship of the craft but shuold be about the integrity and impact of the output. The real challenge is ensuring that what we create with these advanced tools is not only innovative but also responsible and answerable. As we continue to integrate AI into more aspects of our lives, we must focus less on who—or what—authored the content and more on the ethical implications of releasing it into the world. This is where the true ethical discourse lies, and it is here that our attention should be firmly fixed.
Thursday, August 8, 2024
The Cognitive Leap Theory
With the arrival of AI, education is experiencing a profound
shift, one that requires a rethinking of how we design and implement learning
activities. This shift is captured in the cognitive leap theory, which posits
that AI is not just an add-on to traditional education but a transformative
force that redefines the learning process itself. The Cognitive Leap theory is
a core part of a larger AI-positive pedagogy framework.
Traditionally, educational activities have been structured
around original or revised Bloom’s Taxonomy, a framework that organizes
cognitive skills from basic recall of facts (Remember) to higher-order skills
like Evaluation and Creation. While Bloom’s pyramid was often interpreted as a
sequential progression, Bloom himself never insisted on a strict hierarchy. In
fact, with the integration of AI into the classroom, the importance of these
skills is being rebalanced. The higher-order skills, particularly those
involving critical evaluation, are gaining prominence in ways that were
previously unimaginable.
In an AI-positive pedagogical approach, the focus shifts
from merely applying and analyzing information—tasks typically associated with
mid-level cognitive engagement—to critically evaluating and improving
AI-generated outputs. This represents a significant cognitive leap. Instead of
simply completing tasks, students are now challenged to scrutinize AI outputs
for accuracy, bias, and effectiveness in communication. This shift not only
fosters deeper cognitive engagement but also prepares students to navigate the
complex landscape of AI-driven information.
A key component of this approach is the development of
meta-AI skills. These skills encompass the ability to formulate effective (rich)
inquiries or prompts for AI, to inject original ideas into these prompts, and,
crucially, to critically assess the AI’s responses. This assessment is not a
one-time task but part of an iterative loop where students evaluate, re-prompt,
and refine until the output meets a high standard of quality. This process not
only sharpens their analytical skills but also enhances their creative
abilities, as they learn to think critically about the inputs and outputs of AI
systems.
Moreover, the traditional view that learning progresses
linearly through Bloom’s Taxonomy is being upended. In the AI-enhanced
classroom, evaluation and creation are no longer the endpoints of learning but
are increasingly becoming the starting points. Students must begin by
evaluating AI-generated content and then proceed to improve it, a process that
requires a deep understanding of context, an awareness of potential biases, and
the ability to communicate effectively. This reordering of cognitive priorities
is at the heart of the cognitive leap theory, which emphasizes that the future
of education lies in teaching students not just to perform tasks but to engage
in higher-order thinking at every stage of the learning process.
The implications of this shift are serious. Educators must
rethink how they design assignments, moving away from traditional task-based
assessments toward activities that challenge students to evaluate and improve
upon AI-generated outputs. This requires a new kind of pedagogy, one that is
flexible, iterative, and deeply engaged with the possibilities and limitations
of AI.
By reimagining the role of higher-order thinking skills and emphasizing the critical evaluation of AI outputs, we can prepare students for a future where cognitive engagement is more important than ever. This is not just about adapting to new technology; it is about transforming the way we think about learning itself.
Thursday, August 1, 2024
Meet Jinni, a Universal Assistant Bot
Take Dr. Nguyen, for instance. A junior professor with a packed schedule, she was just invited to present at a conference in Milan but wasn't sure how to get funding. She turned to Jinni.
"Good afternoon, Professor Nguyen. What do you need today?" Jinni asked.
"I want to attend a conference in Milan. Can I get support?" she inquired.
It added, "If you’d rather tell me the details about the conference and upload the invitation letter, I can file the request for you. Or, you can follow the link and do it yourself."
Professor Nguyen appreciated the options and the clarity, and chose to upload her details, letting Jinni handle the rest. Within a minute, Jinni said "Done, you shuold hear from the dean's office within a week. I alrready checked your eligibility, and recommended the Associate Dean to approve."
Then there was Mr. Thompson, a new staff member who discovered a puddle in the lobby after a rainy night. He pulled out his phone and described the situation to Jinni.
"You need to file an urgent facilities request. Here’s the link. Would you like me to file one for you? If yes, take a picture of the puddle," Jinni offered. "But if it’s really bad, you may want to call them. Do you want me to dial?"
Mr. Thompson opted for the latter, and within moments, Jinni had connected him to the facilities team.
Finally, there was Jose, a student who had missed the course drop deadline because of a bad flu. Anxious and unsure what to do, he asked Jinni for help.
"Sorry to hear you’ve been sick. Jose. Yes, there is a petition you can file with the Registrar," Jinni replied. "I can do it for you, but I need a few more details. Do you have a note from your doctor? If not, you should get it first, then take a picture of it for me. If you used the Campus Health Center, I can contact them for you to request documentation. I will then write and submit the petition on your behalf. I will also need a few details - which class, the instructore's name, when you got sick, etc." Jose was relieved to find a straightforward solution to his problem and began to answer Jinni's questions one by one.
The technology to create a universal agent bot like Jinni is not yet on the open market, but all elements do already exist as prototypes. More advanced customizable AI models, trained on extensive and diverse datasets, are essential to handle such tasks. More active, agentic AI also does exist. It can file and submit forms, not just find them. But even if we could to simply find and interpret policy and procedures, and point users to the right forms, it would alredy be a huge step forward.
Monday, July 29, 2024
AI is an Amateur Savant
Most people who use AI think it is great in general but believe it does not grasp their area of specialization very well. As an applied philosopher, I create intellectual tools to help others think through their problems. I find AI excellent at clarifying and explaining ideas, but it has never generated an original idea worth writing about. I have yet to see reports from others in any discipline that AI has independently produced groundbreaking ideas.
AI can handle large amounts of data and provide coherent, accurate responses across various fields. This ability is comparable to a well-informed amateur who has a broad understanding but lacks deep expertise. AI can recount historical facts, explain scientific principles, and offer legal insights based on data patterns, yet it falls short in deeper, more nuanced analysis.
In my case, AI can assist by summarizing existing theories or offering possible objections or additional arguments. However, it lacks the ability to generate a genuinely novel idea. I use it a lot, and not even once did it produce anything like that. This limitation stems from its reliance on pre-existing data and patterns, preventing it from achieving the level of innovation that human professionals bring to their fields. Some believe that this limitation will soon be overcome, but I do not think so. It seems to be an intrinsic limitation, a function of AI's way of training.
Professionals/experts, whether in philosophy, medicine, or history, possess a depth of understanding developed through extensive education and practical experience. They apply complex methodologies, critical thinking, and ethical considerations that AI cannot replicate. A doctor considers the patient's history and unique implications of treatments, while a professional historian places events within a broader socio-cultural context. AI, despite its capabilities, often misses these subtleties. It is, in some sense, a savant: a fast, amazing, but inexperienced thinker.
The gap between a capable amateur and a professional/expert might seem small, especially from the point of view of the amateur. However, it is huge and is rooted in the depth of expertise, critical thinking, and the ability to judge that professionals possess; it is a function of intellect, experience, and education. This gap is where educators should look to adapt the curriculum.
In education, we should focus on that gap between the amateur and the professional and conceptualize it as the ultimate learning outcome, then build new skill ladders to claim there. Students need to understand and conquer the gap between AI and a professional expert. These meta-AI skills are our true North. AI can support this learning process by providing clear explanations and diverse perspectives, but it cannot replace the nuanced understanding and innovation that human professionals offer.
Wednesday, July 24, 2024
What percentage of my text is AI-generated?
Go ahead, ask me the question. However, I would in turn ask you to specify which of the following kinds of assistance from AI you are interested in.
- Distilling information into summaries
- Revamping and recasting content
- Polishing grammar, spelling, and punctuation
- Sparking ideas and crafting titles
- Conjuring additional arguments or perspectives
- Spotting potential counterarguments or objections
- Constructing and organizing content
- Juxtaposing points from multiple sources
- Scrutinizing and refining existing content
- Demystifying complex ideas or jargon
- Architecting outlines and organizational structures
- Fashioning examples or illustrations
- Tailoring content for different audiences or formats
- Forging hooks or attention-grabbing openings
- Sculpting strong conclusions or call-to-actions
- Unearthing relevant quotes or citations
- Decoding concepts in simpler terms
- Fleshing out brief points or ideas
- Trimming verbose text
- Honing clarity and coherence
- Smoothing the flow between paragraphs or sections
- Concocting metaphors or analogies
- Verifying and authenticating information
- Proposing synonyms or alternative phrasing
- Pinpointing and eliminating redundancies
- Diversifying sentence variety and structure
- Maintaining consistency in tone and style
- Aligning content with specific style guides
- Devising keywords for SEO optimization
- Assembling bullet points or numbered lists
- Bridging sections with appropriate transitions
- Flagging areas that need more elaboration
- Accentuating key takeaways or main points
- Formulating questions for further exploration
- Contextualizing with background information
- Envisioning visual elements or data representations
- Detecting potential areas of bias or subjectivity
- Inventing catchy titles or headlines
- Streamlining the logical flow of arguments
- Boosting text engagement and persuasiveness
- Rooting out and rectifying logical fallacies
- Imagining hypothetical scenarios or case studies
- Illuminating alternative perspectives on a topic
- Weaving in storytelling elements
- Uncovering gaps in research or argumentation
- Producing counterexamples or rebuttals
- Bolstering weak arguments
- Harmonizing tense and voice inconsistencies
- Composing topic sentences for paragraphs
- Integrating data or statistics effectively
- Devising analogies to explain complex concepts
- Injecting humor or wit
- Eradicating passive voice usage
- Compiling topic-specific vocabulary lists
- Enhancing paragraph transitions
- Untangling run-on sentences
- Articulating thesis statements or main arguments
- Infusing content with sensory details
- Resolving dangling modifiers
- Conceiving potential research questions
- Incorporating rhetorical devices
- Rectifying pronoun inconsistencies
- Anticipating potential counterarguments
- Embedding anecdotes effectively
- Mending comma splices
- Drafting potential interview questions
- Sprinkling in cultural references
- Correcting subject-verb agreement errors
- Designing potential survey questions
- Adorning text with figurative language
- Repositioning misplaced modifiers
- Brainstorming potential titles for sections or chapters
- Integrating expert opinions
- Paring down wordiness
- Exploring potential subtopics
- Weaving in statistical data
- Eliminating tautologies
- Coining potential taglines or slogans
- Embedding historical context
- Untangling mixed metaphors
- Developing potential FAQs and answers
- Incorporating scientific terminology
- Fixing split infinitives
- Generating potential discussion points
- Blending in technical jargon
- Expunging clichés
- Crafting potential calls-to-action
- Inserting industry-specific terms
- Replacing euphemisms
- Extracting potential pullout quotes
- Interweaving mathematical concepts
- Eliminating redundant phrasing
- Compiling potential glossary terms and definitions
- Introducing philosophical concepts
- Standardizing formatting
- Curating potential appendix content
- Incorporating legal terminology
- Clarifying ambiguous pronouns
- Cataloging potential index terms
- Synthesizing interdisciplinary perspectives
- Writing long list of AI uses for content generation
Four Myths About AI
AI is often vilified, with myths shaping public perception more than facts. Let us dispel four common myths about AI and present a more bala...
-
In the ongoing narrative of education's transformation, AI's integration has prompted a profound reassessment of what constitutes un...
-
The notion of measuring "AI-generated text" as a fixed percentage of an academic submission is fundamentally flawed. This metric i...
-
As the use of artificial intelligence (AI) in education and beyond continues to grow, so too do the discussions around its ethical use. Howe...