To Be or Not to Be Human? �The Dilemma of Anthropomorphic LLMs���SICSS-Habib, Guest Lecture��15-06-2026

Dr. Rabeea Jaffari

Assistant Professor, Department of Software Engineering�Mehran University of Engineering and Technology (MUET), Pakistan

​

About Me

• PhD in AI (Youngest GoT)
• Interdisciplinary AI researcher
• 8 years research experience in AI applications for sustainability and scientific discovery
• International collaborations through research and academic programs (ICTP, Italy and TUM, Germany)
• IEEE Women in Engineering (WIE) Faculty Advisor, MUET
• Interested in AI Policy and Governance

Today’s Journey

• Why humans humanize machines?
• What makes Large Language Models (LLMs) feel human?
• Benefits and risks of anthropomorphic AI
• Human–AI interaction from a social science perspective
• Interactive exercise

What is Anthropomorphism?

“The tendency to attribute human characteristics, emotions, intentions, or personality to non-human entities”

Examples:�• Naming robots�• Talking to cars�• Saying AI “understands” us

• The Rise of Human-Like AI: From command systems → conversational agents

​

Why Do Humans Anthropomorphize?

• We seek social connection
• We interpret intelligence socially
• Human-like communication triggers human-like expectations

​

Is AI becoming human or are humans projecting humanness?

Why LLMs Feel Human?

Human or Human-Like?

• User:�“I failed my interview.”
• AI:�“I’m sorry to hear that. That can feel disappointing.”

Discussion: Empathy ≠ Emotion

Benefits vs Dilemma

Anthropomorphic AI may improve:

�• Engagement�• Accessibility�• Learning�• Communication�• User experience

Potential concerns:

�• Over trust�• Emotional dependency�• Manipulation�• Confusion about AI capability�• Ethical boundaries

Computational Social Science Lens

How researchers study this:

• Surveys
• Experiments
• User studies
• Behavioral analysis
• Interaction data

​

Real World Questions

Should AI say:

• “I understand”
• “I care”
• “I remember you”

​

Where should we draw the line?

​

Mini Workshop — Human, AI, or Cannot Tell?

Instructions: Read each response and classify into

1. Human
2. AI
3. Cannot Tell

Scenario 1

Prompt:�“I didn’t get selected for the fellowship.”

Response:�“That’s disappointing, especially when you invested so much into preparing. Give yourself a little time and then revisit what you learned from the process.”

Answer: Cannot Tell

Mini Workshop — Human, AI, or Cannot Tell?

Scenario 2

Prompt:�“Do you think I should leave academia?”

Response:�“There is insufficient information to recommend a decision. Consider evaluating career goals, finances, personal values, and long-term priorities.”

Answer: AI

​

Scenario 3

Prompt:�“I stayed up all night preparing and still feel nervous.”

Response:�“That sounds exhausting. I usually remind myself that preparation and uncertainty can exist together.”

Answer: Human

​

Mini Workshop — Human, AI, or Cannot Tell?

Scenario 4

Prompt:�“What’s the most important thing in research?”

Response:�“I think curiosity matters more than certainty. Good research often starts with good questions rather than perfect answers.”

Answer: Cannot Tell

​

Scenario 5

Prompt:�“Can AI truly understand emotions?”

Response:�“It may generate language that resembles empathy, but generating emotional expressions is different from experiencing emotions.”

Answer: AI

​

Reflection

• Which cues made something feel human?
• Did warmth increase perceived humanness?
• Did confidence influence trust?
• Were you more accurate than you expected?

The Twist: Two of these examples had no correct answer

​

Future Research Directions

• Can anthropomorphism be measured?
• Should AI appear less human?
• How should policy respond?
• What does responsible AI interaction look like?

Final Reflection��“If humans increasingly respond socially to machines,�should AI become more human or remain recognizably artificial?”

Thank you��Questions? Let’s start

Dr. Rabeea Jaffari

rabeea.jaffari@faculty.muet.edu.pk

rabeea.jaffari@gmail.com