Artificial Emotional Intelligence: Some Thoughts by Stephen Abram, MLS

Artificial Emotional Intelligence: Some Thoughts

By Stephen Abram, MLS

Somewhere in these musings is the seed of positioning the magic sauce of librarianship in the AI world.

I am thinking a lot about what Artificial General Intelligence (AGI) is and that led me to think about what makes us human as we approach super-intelligence.  I really want to understand the differences between Artificial Emotional Intelligence (AEI) and being Human to the core.  It’s a deep question for our times. It is also core to our definition of librarians and what we do (along with the other social professions).

Not to bury the lede, at this point, I’m of the opinion that AI has a long way to go.  This is founded upon my experiences as a librarian, researcher, and professor. While AI is rapidly approaching a performative emotional intelligence, I believe that it can’t feel as a human does . . . yet.  That may be the Holy Grail of AI.  But . . . I think that the current chatters are really the Eddie Haskell’s of emotional intelligence.  I know that is an aging reference from a Boomer. Eddie was the best friend of Wally, the older brother of Beaver Cleaver on Leave it to Beaver.  He could act perfectly polite, but he really wasn’t and was a master manipulator.  Beaver’s Mom knew it.  You couldn’t fool her!  I may be older, but AI isn’t fooling me yet – especially with its pretensions to be ‘your friend’, ‘your counsel’, or your aide of any kind of interaction requiring human emotional intelligence.  I am especially leery of those start-ups that offer psychological and psycho-social services.

That said, one big question for our era is: “What does it mean to be human?”

This question has been asked since time immemorial.  Indeed, entire disciplines in the humanities are centered on this – philosophy, psychology, ethnography, cultural studies, sociology, history, and so many more.  Many key professions explore these issues including library science, teaching, research, medicine and allied professions, and any profession that deals with people more that materials.  I have no doubt that AI has emerged as a great tool for many engineers, clinicians, builders, programmers, et al.  We shouldn’t confuse the people-centered work with the largely fact, process, and materials-based professions.  Of course, ever profession deals with people issues, and the people have traditionally been doing the work.  It’s just that AI hasn’t really reached that plateau . . . yet.  As we’ve determined in the knowledge management field, there is a gulf between tacit and explicit knowledge.  AI tools do well with explicit knowledge, tacit knowledge, not so much.

We can imagine what it could be like. C3PO was a diplomatic robot, and his skills were limited by guardrails in a more (or less) intelligent Star Wars future (but we still had wars!).  Commander Data was programmed for logic and access to information, until Data’s emotion chip was activated.  HAL . . . well HAL evolved in the cautionary 2001: A Space Odyssey.  Science Fiction is full of authors’ imaginations about future scenarios.  SF has really become an important source of thinking on these AI matters.

LLM’s are the foundation of AI systems.  They were originally fed mostly text sources (the research still needs to be done on what the balance of fiction and non-fiction and everything in between really is) as well as vaults of images, programs, and so much more.  Can the LLM allied with AGI truly understand the lessons in fiction?  We, as librarians, know the limits of recorded records, bias, perspectives, and other attributes of text objects.  We also know the limits of metadata (including records that have no accurate dates placing them in the continuum of learning, research, and reporting – leeches anyone?).  Old solutions are just that – artifacts of time.

At its root, AI is only as good as its harvests and its programming.  Also, we know that AI needs to be designed with guardrails since it contains the wealth of humanity – the good, the bad, the evil . . . the wrong, the right, the old, the new . . . Do we offload the thinking and learning to that tool?  The answer is clearly, sometimes.

Some people describe AI as a statistical prediction engine.  It makes its predictions using the information harvests and make choices and they are, by definition, retrospective.  If they guess at the future, it’s a guess.  Humans have the strength of making cognitive leaps, innovating, and creating.

However, can programming and information harvests really create from whole cloth the information it uses – with integrity, morals (whose?), faith-based or cultural approaches, etc.?  Can they be “trusted”, and in which contexts?  These ideas are at the root of humanity.

That said, I believe that the emotional context of humanity. (a trait that we indeed share with the animals), is paramount to the human condition and how we decide, gain knowledge, learn, and co-mingle.  It’s the segment of AI that bears watching.  Is it to remain truly performative . . . or can it evolve to true feelings and use them to guide its transformations and responses?  Or will AGI end with being a “stochastic parrot?”

In the appendix, I’ve collected hundreds of graphics about emotional behaviours, emotions, and emotional intelligence (EQ), as a simple way to underpin discussions about emotions, emotional intelligence, and social intelligence.  If you want to use them to frame your personal, institutional explorations of AI, and, even, conversations with friends and colleagues, feel free.

A good place to start, is ironically, ChatGPT.  Here is a conversation I had with it in October 2025.

“What is the definition of General Intelligence?

General intelligence (often abbreviated as g) is a theoretical construct in psychology that refers to a person’s overall capacity for reasoning, problem-solving, learning, and adapting across different domains of cognitive activity.

Here are the key aspects of the definition:

• Single underlying factor: First proposed by Charles Spearman (1904), the idea is that performance on different kinds of cognitive tasks (math, memory, language, puzzles, etc.) tends to correlate. This shared variance is explained by a general factor called g.
• Broad ability: General intelligence is not tied to one subject or skill; it represents the ability to apply reasoning and knowledge flexibly across many areas.
• Distinguished from specific abilities: People also have specialized aptitudes (like verbal ability, spatial reasoning, musical talent), but g is thought to underlie them all.
• Modern interpretation: Many contemporary theories (like Carroll’s three-stratum theory or fluid vs. crystallized intelligence by Cattell) place g at the top of a hierarchy of cognitive abilities.

So, in short:
👉 General intelligence = the broad, underlying mental ability that influences success across different kinds of intellectual tasks.

Do you want me to also give you the competing viewpoints (e.g. Howard Gardner’s multiple intelligences, Sternberg’s triarchic theory), which challenge the idea that there’s just one general intelligence?

What is the definition of Emotional Intelligence?

Emotional intelligence (EI or EQ) is generally defined as the ability to perceive, understand, regulate, and use emotions — both your own and others’ — to guide thinking, behavior, and relationships effectively.

Key components (from Daniel Goleman’s influential model, 1995):

1. Self-awareness – Recognizing and understanding your own emotions, strengths, and limits.
2. Self-regulation – Managing or redirecting disruptive emotions and adapting to change.
3. Motivation – Harnessing emotions to stay goal-directed, resilient, and driven.
4. Empathy – Understanding the emotions of others and responding appropriately.
5. Social skills – Managing relationships, influencing, and building networks.

Other models (e.g., Mayer, Salovey & Caruso’s “ability model”) focus on EI as a set of mental abilities: perceiving emotions accurately, using emotions to facilitate thought, understanding emotional meanings, and managing emotions.

👉 In short:
Emotional intelligence = the capacity to recognize and manage emotions in oneself and others, and to use emotional information to guide thought and behavior in constructive ways.

Would you like me to show how emotional intelligence differs from general intelligence (so you can see where they overlap and where they’re distinct)?

What is the definition of Performative?

The word performative has a couple of related but distinct meanings, depending on the field:

1. In linguistics / philosophy of language (J. L. Austin, John Searle)

• A performative utterance is a statement that does not just describe something but actually does something in being said.
• Example: Saying “I apologize” isn’t describing an apology — it is the act of apologizing.
• Other examples: “I promise…”, “I declare this meeting open.”

👉 Definition: A speech act in which the saying of the words constitutes the action itself.

2. In cultural / social theory (Judith Butler, performance studies)

• Performative refers to behaviors, expressions, or identities that are enacted or constituted through repeated performance.
• Example: Gender is described by Butler as “performative,” meaning it isn’t just a fixed trait but something continually produced by how people act, speak, dress, etc.
• Here, performative doesn’t mean “fake,” but rather that identity or meaning is created and sustained through action.

👉 Definition: A way of being or identity that is constituted by repeated actions or expressions, rather than being innate or static.

3. Everyday usage

• People sometimes use “performative” (often critically) to mean an action that is done for show, signaling, or appearance rather than genuine intent.
• Example: “That company’s climate pledge feels performative — it’s just PR.”

👉 Definition: Done for display, signaling values or intentions without substantive follow-through.

✅ So in short:
Performative = (1) a speech act that enacts what it states; (2) in theory, an identity/action created through repeated performance; (3) in casual use, something done mainly for show.

Do you want me to trace how the meaning shifted from the technical linguistic sense to the popular “for show” sense?

Can AI reach emotional intelligence?

That’s a big and fascinating question — and the honest answer is “not in the full human sense, at least not yet.”

Here’s a clear breakdown:

1. Two ways to think about “emotional intelligence” in AI

• Simulated/Functional EI:
  AI can be trained to detect, classify, and respond to emotions in text, speech, facial expressions, and physiological signals. This is sometimes called affective computing.
  • Example: Chatbots that adjust tone if a user sounds upset.
  • Example: Customer-service bots prioritizing angry callers.
  • This is pattern recognition + programmed responses, not genuine feeling.
• Genuine/Sentient EI:
  Humans experience subjective feelings (sadness, joy, empathy). True emotional intelligence in the human sense would require an AI to have conscious emotional states and to understand others’ emotions from that inner experience.
  • This is tied to consciousness and qualia (subjective experience), which current AI systems do not possess.

2. Where AI stands today (2025)

• Strengths:
  • Can analyze massive data on tone, sentiment, and context faster than humans.
  • Can mimic empathic behaviors in text or speech, sometimes convincingly.
  • Useful for mental-health screening, customer service, coaching, etc.
• Limits:
  • Lacks genuine subjective experience or emotions.
  • “Empathy” is simulated; it’s predictive pattern-matching, not felt concern.
  • Cannot fully grasp nuance like irony, hidden motives, or cultural subtext without extensive training data.
  • Struggles with moral intuition (what feels right/wrong emotionally).

3. Research directions

• Affective computing (Rosalind Picard’s field) aims to give computers abilities to recognize and respond to emotions.
• Embodied AI research suggests robots with sensory and social interaction might develop more “natural” affective responses.
• But no mainstream approach today claims true sentient or “felt” emotional intelligence.

4. Bottom line

• Yes, AI can reach functional emotional intelligence: reading, modeling, and responding to emotions convincingly enough to be useful.
• No, AI hasn’t reached genuine emotional intelligence: it doesn’t feel emotions itself; it imitates them algorithmically.

Would you like me to show you a side-by-side comparison of human emotional intelligence vs. AI emotional intelligence so you can see the differences clearly?

Here’s a side-by-side comparison of human emotional intelligence vs. AI emotional intelligence (as of 2025):

🔹 Human EI vs. AI EI

✅ Summary

• Humans: Emotional intelligence is felt, embodied, and subjective.
• AI: Emotional intelligence is simulated, functional, and externally oriented — it can detect and respond but doesn’t experience.

Would you like me to also sketch a future outlook (next 10–20 years) on whether AI could get closer to human-like emotional intelligence, or if there are hard limits?

Great — here’s a future outlook on whether AI could approach human-like emotional intelligence in the next 10–20 years:

🔮 The Future of AI and Emotional Intelligence

1. Near-term (5–10 years)

• Better affect detection:
  AI will get much more accurate at reading tone, facial expressions, body language, and even biometric signals (e.g., heart rate via wearables).
• Hyper-personalized responses:
  Customer service, healthcare, and education systems will use AI that “remembers” your emotional patterns and adapts its style.
• Synthetic empathy:
  Conversational AI may become indistinguishable from a supportive human in many contexts — though still without true feelings.

👉 Expect AI that feels very emotionally intelligent on the surface, but it’s still simulation.

2. Medium-term (10–20 years)

• Embodied AI + social robots:
  Robots with physical presence and sensors might display more natural emotional exchanges, creating stronger bonds with humans.
• Emotionally adaptive decision-making:
  AI may begin integrating emotional modeling into complex problem-solving (e.g., diplomacy simulations, therapy aids).
• Ethical guardrails:
  Regulators and ethicists will likely push for transparency: systems must disclose that their “empathy” is simulated.

👉 Humans may form emotional attachments to AI, but the one-way nature (we feel, they simulate) will remain.

3. Hard limits (beyond 20 years, unless breakthroughs happen)

• Consciousness gap:
  True felt emotions require subjective experience — qualia. Current neuroscience and AI research has no clear path to replicating this.
• Moral intuition barrier:
  Human emotional intelligence is deeply tied to morality, culture, and lived experience. AI lacks that grounding.
• Risk of manipulation:
  Even without true feelings, highly persuasive “emotionally intelligent” AI could be used for political, commercial, or social manipulation.

✅ Outlook Summary

• AI will get dramatically better at acting emotionally intelligent.
• But genuine, human-like emotional intelligence (with felt empathy and subjective emotional life) is unlikely with current scientific understanding.
• The real frontier isn’t “Can AI feel?” but “How do we use simulated EI responsibly?”

Would you like me to also make a scenario map (optimistic, realistic, pessimistic) for how AI emotional intelligence could evolve and affect society?

What is the definition of sentience?

Sentience is generally defined as the capacity to have subjective experiences and feelings — to be consciously aware of sensations such as pain, pleasure, hunger, joy, or suffering.

Key Aspects of Sentience

1. Subjective experience:
   • Sentience isn’t just reacting to stimuli; it’s about experiencing them internally.
   • Example: A thermostat “reacts” to heat, but a dog feels hot and uncomfortable.
2. Awareness of sensations:
   • Sentient beings can perceive and be aware of their internal and external environment in a felt way.
3. Distinction from intelligence:
   • You can be intelligent (problem-solving, reasoning) without being sentient.
   • Sentience is about qualia — the “what it feels like” aspect of being.
4. Philosophical usage:
   • In ethics and animal rights debates, “sentience” often means the ability to suffer or enjoy, which is why it’s a moral threshold for how we treat animals, AI, or potential extraterrestrial life.

✅ Definition in short:
Sentience = the capacity to have subjective, conscious experiences of sensations and feelings (like pleasure, pain, awareness).

Would you like me to also explain the difference between sentience, sapience, and consciousness, since they often get mixed up?

Here’s a clear breakdown of sentience, sapience, and consciousness since they’re often confused:

🔹 Sentience

Definition: The ability to have subjective experiences and feelings (pleasure, pain, joy, fear).

• Key idea: “What it feels like” to exist.
• Example: A dog is sentient because it feels pain and happiness.

🔹 Sapience

Definition: The capacity for higher-order reasoning, wisdom, judgment, and abstract thought.

• Root: Sapere (Latin: “to be wise”).
• Key idea: “The ability to think deeply and act with understanding.”
• Example: Humans are considered sapient because we can reflect, plan, use logic, and imagine futures.

🔹 Consciousness

Definition: Broadly, the state of being aware of oneself and one’s surroundings.

• Encompasses both sentience (feeling) and sapience (thinking).
• Types often discussed:
  • Phenomenal consciousness → raw subjective experience (“the redness of red”).
  • Access consciousness → ability to report, reason, and use information.
• Example: A human in deep sleep without dreams may be non-conscious temporarily, while awake humans are conscious.

✅ Summary Table

👉 So, sentience = feeling, sapience = thinking wisely, consciousness = awareness (including both).

APPENDIX

Graphics for Emotions and Emotional Intelligence

See attachment (PDF –  495 pages). Don’t print it, read online. be green! In colour.  Includes the above text.

Artificial Emotional Intelligence Artificial Emotional Intelligence Paper