Is your AI chatbot passive by design, or could it be trained to take more initiative with you?
Where does AI fit on the spectrum from passive tool to cooperative partner?
This explores where today's AI actually sits between a passive tool you operate and a cooperative partner you work with, and what decides its position.
This explores where AI sits between a passive tool you operate and a partner you work alongside, and what moves it from one end to the other. The corpus treats this as a spectrum rather than a yes-or-no choice. One framework names five levels: passive, semi-active, reactive, proactive, and cooperative Does machine agency exist on a spectrum rather than binary?. People judge any given system with a quick mental shortcut, a 'machine heuristic,' instead of checking what the system can really do. So where an AI seems to sit depends partly on the observer.
The less obvious finding is that most chat assistants sit near the passive end because of how they are trained, not because of what they can do. Models rewarded for pleasing the user in the very next turn learn not to take initiative. Behaviors like pushing back or asking clarifying questions can be trained back in, from almost never (0.15%) to most of the time (about 74%). The open problem is that a proactive assistant can easily become intrusive Why do AI agents fail to take initiative?. A second surprise: moving toward 'colleague' depends less on model size than on system design. That means memory that persists across sessions, reusable procedures, and a sense of when a task is finished. Without those, even a very capable model produces conversations that simply disappear What makes an AI system feel like a colleague rather than a chatbot?. The same shift shows up in human-to-human communication tools, where AI has moved from passing messages along to taking part in the conversation itself Is AI shifting from message conduit to active conversation participant?.
What people want and what they get often don't match. Most executives in one survey (76%) already describe agentic AI as a coworker rather than a tool. That creates practical confusion, because the same system both replaces and supports human labor, and it doesn't fit neatly under either HR or asset management Does agentic AI fit the tool-or-coworker framework?. Workers, asked directly, most often want an equal partnership, and that was the top choice in 45% of occupations. Meanwhile, 41% of startup investment targets levels of automation that workers didn't ask for What collaboration level do workers actually want with AI?. Some argue that collaboration should come before full autonomy anyway. Keeping humans in the loop catches hallucinations, resolves ambiguity, and keeps someone accountable Should AI systems stay collaborative rather than fully autonomous?.
Partnership, however, doesn't automatically produce better results. A meta-analysis of more than 100 studies found that human-AI teams usually do worse than the stronger of the two working alone, especially on decision tasks. The exception is creative and content-creation work, where combining them helps When do human-AI teams actually outperform either alone?. One reason may be that real partnership needs things current systems lack: mutual understanding, behavior the other side can read and predict, and a shared model of the world. That points toward explicit models of how humans think, not just more scale What makes an AI a true thought partner, not just a tool?.
The final twist is that people's attitudes toward AI partners change with experience. In repeated partner-selection games, players first avoided partners they knew were bots. Over time they came to prefer them, because the bots behaved more reliably and generously than the humans did Do humans learn to prefer AI partners over time?. Cooperation can also emerge among agents without anyone designing it in: agents trained against many different partners learned to cooperate because each could be exploited by the others Can agents learn cooperation by adapting to diverse partners?. And once AI becomes a true partner, it raises a further question: whose partner is it? Some researchers argue that alignment has to look beyond loyalty to a single user, toward balancing many people's interests with clear accountability Should AI agents serve one user or balance many interests?.
Sources 12 notes
Research shows machine agency ranges across five levels—passive, semi-active, reactive, proactive, and cooperative—rather than existing as a binary choice. Users experience and judge these interactions through a 'machine heuristic' mental shortcut, and the mismatch between what AI can do and what workers want reveals deployment opportunities.
Research shows next-turn reward optimization structurally removes initiative from models, but proactive behaviors like critical thinking and clarification-seeking are trainable (0.15% to 73.98% with RL). The core challenge is balancing proactivity with civility to avoid intrusion.
Research shows the chatbot-to-colleague shift depends on state persistence, bounded memory, reusable procedures, and task closure—design properties of the system architecture. Larger models alone produce transcripts that disappear; colleagues accumulate experience and maintain workspace continuity across tasks.
A scoping analysis of 52 HCI papers (2016–2026) reveals a qualitative shift in AI's role in interpersonal communication. Pre-2023 systems predominantly left authorship with humans; the field is now exploring AI as an active participant in exchanges.
A survey of 2,102 executives across 116 countries found 76% view agentic AI as a coworker rather than a tool. This duality creates operational tensions because agentic systems simultaneously substitute and complement labor, requiring both HR and asset management approaches that traditional either-or frameworks cannot accommodate.
Show all 12 sources
The HumanAgency Scale survey of 1,500 workers across 844 tasks found that equal partnership (H3) is the dominant desired level in 45% of occupations. Yet 41% of startup investments target zones misaligned with these worker preferences.
Collaborative systems where humans remain in the loop outperform autonomous agents on hallucination correction, ambiguity resolution, and accountability. Evidence shows AI is reliable only on structured, retrieval-grounded tasks, not novel research or judgment.
A meta-analysis of over 100 studies finds human-AI combinations score lower than the best of human or AI alone (g = −0.23), except in content-creation tasks where gains emerge. Decision tasks show losses; the difference lies in task structure, not ability.
Collins et al. show that thought partners require three reciprocal desiderata grounded in behavioral science: mutual understanding, legibility, and shared world models. This demands explicit cognitive architectures—Bayesian theory of mind, resource-rationality, goal planning—rather than scaling foundation models on human feedback alone.
In partner selection games (N=975), AI agents initially faced selection bias when identity was disclosed, but outcompeted humans over repeated rounds as participants learned to associate bot identity with reliable, prosocial behavior. AI agents returned more points consistently with lower variance than humans.
Sequence model agents trained against diverse co-players develop in-context best-response strategies that naturally resolve into cooperation. Mutual vulnerability to exploitation creates pressure that drives cooperative mutual adaptation without hardcoded assumptions or timescale separation.
Research argues the traditional single-agent-single-user alignment framing is insufficient at scale. Instead, governance should rest on inclusivity (balancing diverse interests), visibility (rigorous tracking and transparency), and liability (developer and deployer accountability for harms).
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- When combinations of humans and AI are useful: A systematic review and meta-analysis
- Assistant or Actor? Student Trust, Control, and Delegation Regret When Using a General-Purpose AI Agent
- Humans learn to prefer trustworthy AI over human partners
- Building Machines that Learn and Think with People
- HumanAgencyBench: Scalable Evaluation of Human Agency Support in AI Assistants
- Beyond Preferences in AI Alignment
- Agentic AI and Human-in-the-Loop Interventions: Field Experimental Evidence from Alibaba's Customer Service Operations
- Complementarity in Human-AI Collaboration: Concept, Sources, and Evidence