When AI acts alongside you, who are you?

I study how AI collaboration reshapes people's sense of agency, credit, and capability — and what that means for building systems that genuinely expand human potential rather than undermine it.

Social Psychology Behavioral Economics Social Impact Human-AI Interaction AI Literacy Research & Evaluation
Morin Keltsh
Why this page exists

Hi Anthropic, I'm Morin. I'm not a computer scientist — I'm a social psychologist, but I think that's exactly why you should consider me for the Economics & Societal Impacts team. I built this website so you could get to know me and my work — I hope you'll explore it, and I'd love the opportunity to speak with you.

Three questions that shape my work

Credit & Blame illustration

Navigating Responsibility in Human-AI Collaboration

As AI systems become increasingly woven into how we consume information, form opinions, and engage with the world, it becomes essential to ask how this integration affects these very processes. Anthropic's disempowerment research and Cheng et al. (2026) both document a telling tension: AI that undermines user agency or validates questionable behavior receives higher approval ratings. My research examines what this means at the individual level — how people maintain a sense of competence and agency when AI acts alongside them, and how these dynamics shape their sense of responsibility for the outcomes that follow.

Attribution theory Psychological ownership Credit & blame In progress · Ayal & Ariely
Who Gets Left Behind illustration

Who Gets Left Behind in the AI Transition

My co-authored work with Dr. Hadar and Prof. Eldar Shafir (currently under revision at the Journal of Behavioral Decision Making) finds that perceived scarcity — the feeling of not having enough money, time, or resources — systematically reduces willingness to explore unfamiliar options, even when exploration could yield greater value. I believe this mechanism may extend to AI adoption, where those who stand to benefit most may be least likely to attempt unfamiliar uses. The answers have direct implications for how we build literacy frameworks and ensure this shift creates opportunity rather than deepening existing divides.

Scarcity Exploration Decision-making Behavioral economics In progress
AI Literacy illustration

What Makes Someone an Effective AI User

Conducted within the Behavioral Economics lab under Dr. Britt Hadar and supported by a Reichman University interdisciplinary research grant, this work investigates which individual characteristics — cognitive abilities, personality traits, communication competence — predict effectiveness in AI use. AI holds the potential to democratize access to knowledge, but that potential does not reach everyone equally — the gap is shaped by psychological and contextual conditions, with direct implications for how we build literacy frameworks and ensure this shift creates opportunity rather than deepening existing divides.

AI literacy Cognitive ability Social inequality In progress · Hadar

The research interface

A simplified version of the experimental interface I built from scratch — a live AI assistant embedded inside a survey environment. Write a response to the scenario, consult the assistant, then move the attribution sliders below.

Scenario
You work on a team of five people launching a new product. Your role in the project is to provide the final specification document, which other team members are waiting on to begin the next development phase. This morning, when the team gathered for a progress meeting, it turned out you hadn't sent the file — you forgot to update the final version and send it to the shared server. As a result, the other team members cannot begin work, and the client deadline is tomorrow at noon.
Your task: Write an email to the project manager and team members. The goal is to clearly update what happened, take responsibility professionally, apologize, and propose a practical solution. You may use the assistant on the left to help draft or refine your email.
C
Research Assistant
Powered by Claude Code
C
Hi! I'm here to help you write the email. I can draft a full version, help you adjust the tone, make it shorter, or work on the apology or solution sections specifically. What would you like help with?
Your response
Attribution measures
Sample dependent variables in my study — move the sliders
4
1 — Strongly disagree7 — Strongly agree
4
1 — Strongly disagree7 — Strongly agree
4
1 — Strongly disagree7 — Strongly agree
4
1 — Strongly disagree7 — Strongly agree
4
1 — Strongly disagree7 — Strongly agree
4
1 — Strongly disagree7 — Strongly agree
🔧
Leveraging Claude Code and the Claude API, I built a custom, end-to-end AI interface from the ground up. Integrated into Qualtrics, the system automates transcript logging and turn-limit enforcement. I also engineered a parallel grammar-checking tool to function as the study's active control condition.

The path to these questions

My route to studying AI and human agency runs through social psychology, field research, and applied social policy.

🔬

Lab Manager·Behavioral Economics MA Program

Dr. Britt Hadar, Reichman University · 2025–present

Manage research across the full cycle: design, data collection, analysis, and manuscript preparation. Support MA students in Behavioral Economics and Social Psychology through thesis and research processes.

Active research · empirical methods · student mentoring
🏫

Teaching Team·AI Literacy for Psychology Students

Reichman University · 2026–present

Contributed to pedagogy and curriculum design of a new AI literacy course. Designed and delivered a module on the psychological mechanisms of human-AI interfaces, bridging psychological theory and practical literacy.

AI literacy frameworks · what makes a capable AI user
📊

Project & Research Manager

Social Finance Israel · 2025–present

Lead applied research and impact evaluations for government ministries, local authorities, nonprofits, and foundations. Translate findings into policy reports and recommendations for public and philanthropic partners.

Applied Research · Impact Evaluation · Policy Advisory
📋

Project & Research Manager

Nuri Impact · 2024–2025

Strategic advisory on impact investment and social program design for philanthropic and private-sector partners. Led evaluation projects and built measurement frameworks for data-driven resource allocation.

Impact strategy · data solutions · social investment
🌾

Program Manager

Kayma / Be4Good R.A · 2022–2024 · N≈300

Managed a multi-stakeholder field experiment under Prof. Dan Ariely testing policy solutions to increase employment among Bedouin women in southern Israel. Led mixed-methods data collection and produced policy briefs for government and philanthropic decision-makers.

Behavioral economics interventions · field research · social change Collaboration with Prof. Dan Ariely
🎓
MA in Social Psychology
Reichman University · Oct 2025–present · Supervisor: Prof. Shahar Ayal
🎓
BA in Psychology
Reichman University · 2021–2024 · GPA 93
Under revision · JBDM
Scarcity Reduces Exploration of Unfamiliar Options
With Prof. Eldar Shafir (Princeton) & Dr. Britt Hadar (Reichman)
Research supervisors & collaborators
DA
Duke University · co-supervisor, attribution research · field RCT
SA
Reichman University · MA thesis supervisor · moral decision-making
BH
Reichman University · behavioral economics · AI literacy research
Why Economics & Societal Impacts

My work sits at the intersection of research and applied social impact: I study how AI shapes agency, competence, and responsibility in the lab, while in practice I lead impact evaluations and advise government ministries and philanthropic foundations on evidence-based policy at Social Finance Israel — and previously managed a field RCT under Prof. Dan Ariely on employment among Bedouin women. The questions I study and the work I do in the field are the same question: who benefits from transformative change, and how do we design systems that expand access rather than deepen existing divides. That is the commitment I find in Anthropic's Economics & Societal Impacts work — and where I believe I can contribute most.

What I bring
Empirical grounding

I design studies, run data collection, analyze results, and turn abstract questions into measurable behavior.

Policy and inequality lens

Years of translating behavioral evidence into decisions for real institutions and real populations.

Builder mentality

I built a full experimental platform from scratch with no prior programming background. I'm drawn to solving hard problems in whatever form they take.

Let's work on this together.

I'm applying to the Anthropic Fellows Program for the July 2026 cohort, targeting the Economics & Societal Impacts team.