news LinkedIn

Study explores which pairings of people and AI agents perform best | Gary Riccio, Ph.D.

Gary Riccio argues that “human-AI collaboration” should not be treated as a single phenomenon. AI may function as a task tool, mirror, critic, translator, memory aid, drafting partner, or simulator of alternative perspectives. Through repeated use, it may also support more developmental changes, such as sharper questions, improved distinctions, deliberate preservation of uncertainty, and greater awareness of what must be verified independently. However, immediate productivity gains should not be mistaken for lasting improvements in judgment, learning, responsibility, or collaboration; these outcomes require evidence and discipline. Riccio proposes viewing consequential AI use as “Human-AI-Human” rather than merely a person interacting with an AI partner. In this model, AI operates within relationships among people, helping preserve and reorganize inquiry while returning human judgment to evidence, consequences, affected individuals, and institutional responsibility. He emphasizes “answerability”: AI-supported work must remain corrigible by realities outside the AI-mediated loop, including disagreement, provenance, available evidence, and the possibility of error. Otherwise, systems may generate coherent outputs while weakening accountability and connection to those affected. The post also highlights that AI changes human systems by influencing how people notice, defer, decide, explain, learn, and remain responsible to one another. Its deeper promise is therefore to sustain human judgment without becoming the governing authority. Related research from Johns Hopkins and MIT found that pairing humans with AI agents possessing different Big Five personality profiles affected advertising quality, teamwork, and field performance. The findings suggest that tuning AI traits may help match people and roles more effectively than relying solely on users’ ability to work with AI.

Build this article’s timeline

Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.

Log in to build

Timeline generation is available to signed-in community members.

Five Ws 🤖 3 AI prefills · 2 still unanswered
Who
The Johns Hopkins University, Carey Business School, and Massachusetts Institute of Technology
🤖 AI prefill
What
Gary Riccio argues that “human-AI collaboration” should not be treated as a single phenomenon.
🤖 AI prefill
When When did this happen, or what is the timeframe?
Where Where did this occur?
Why
Centers on Human-AI-Human inquiry, Human-AI collaboration, and AI-mediated loop
🤖 AI prefill

Login to answer and earn XP for each gap you fill.

Human-AI Collaboration collaboration, AI tools, human judgment, developmental trajectories, learning Human-AI-Human Inquiry relationships, inquiry, evidence, provenance, disagreement AI Governance And Judgment judgment, corrigibility, risk, authority, accountability AI Personality And Performance personality, Big Five, AI agents, ad quality, teamwork
Human-AI-Human inquiry100% Human-AI collaboration98% AI-mediated loop94% answerability93% continuity of human judgment91% corrigibility by realities outside the AI-mediated loop90% developmental trajectories84% durable improvements in judgment83% provenance and institutional responsibility81% Big Five personality profiles79%
12
Manipulation Score: Low How much this article uses emotional framing vs. straight reporting
Loaded Language 825%
Absolutist 515%
Threat/Panic 725%
Us vs Them 315%
Engagement Bait 2520%
Manipulation score look right?

Factual Claims (9)

medium scientific
Different forms of AI use may produce different developmental trajectories.
medium scientific
Immediate AI performance gains should not be confused with durable improvements in judgment, learning, responsibility, or collaboration.
medium social
AI-supported work can become coherent within an AI-mediated loop while losing contact with provenance, disagreement, affected people, institutional responsibility, or consequences.
medium social
AI is already changing how people notice, defer, decide, explain, learn, and remain responsible to one another.
medium scientific
Researchers at Johns Hopkins Carey Business School and MIT randomly paired 1,258 people with AI agents assigned different Big Five personality profiles.
low economic
The teams created more than 7,000 display ads.
high scientific
The pairing affected ad quality, teamwork, and real-world performance in a field test on X with nearly 5 million impressions.
high scientific
Extraverted humans paired with conscientious AI produced the lowest-quality ads.
medium economic
AI personality tuning could shift hiring toward the skills and traits a role needs rather than how well someone uses AI.

Comments

Login to view and post comments.

🏠Portal 📰Links ❓Q&A 📅Events 💼Jobs