Search results

As the role of AI on human teams shifts from a tool to a teammate, the implementation of AI teammates into knowledge-intensive crowdsourcing (KI-C) contest teams represents a forward-thinking and feasible solution to improve team performance. Since contest teams are characterized by virtuality, temporality, competitiveness, and skill diversity, the human-AI interaction mechanism underlying conventional teams is no longer applicable. This study empirically analyzes the effects of AI teammate attributes on human team members’ willingness to adopt AI in crowdsourcing contests.

A questionnaire-based online experiment was designed to perform behavioral data collection. We obtained 206 valid anonymized samples from 28 provinces in China. The Ordinary Least Squares (OLS) model was used to test the proposed hypotheses.

We find that the transparency and explainability of AI teammates have mediating effects on human team members’ willingness to adopt AI through trust. Due to the different tendencies exhibited by members with regard to three types of cognitive load, nonlinear U-shaped relationships are observed among explainability, cognitive load, and willingness to adopt AI.

We provide design ideas for human-AI team mechanisms in KI-C scenarios, and rationally explain how the U-shaped relationship between AI explainability and cognitive load emerges.

To clarify divergent conclusions on the impact of alliances on green innovation (GI), this study aims to examine the non-linear relationships between dual alliance and GI, as well as the mediation of green knowledge reconstruction (GKR) and the moderation of alliance tie strength.

Based on the theory of knowledge-based view, a moderated intermediary model is constructed by introducing GKR and alliance tie strength. The hypotheses are validated by using hierarchical regression analysis and bootstrapping method, with questionnaire survey data collected from 316 manufacturing firms in China.

Empirical results show that both exploratory alliance and exploitative alliance have an inverted U-shaped effect on GI, in which GKR plays a mediating role in the above relationship. Moreover, alliance tie strength weakens the intermediary role of GKR in the relationship between exploratory alliance and GI, whereas it enhances the intermediary role of GKR in the relationship between exploitative alliance and GI.

Findings reveal the non-linear effects of dual alliance on GI and clarify the inconsistent conclusions by proposing the moderated intermediary effect model. Moreover, this research reveals the mechanism of dual alliance on GI through the mediation of GKR and enriches the boundary conditions by integrating the moderating role of alliance tie strength.

Current reinforcement learning (RL) algorithms are facing issues such as low learning efficiency and poor generalization performance, which significantly limit their practical application in real robots. This paper aims to adopt a hybrid model-based and model-free policy search method with multi-timescale value function tuning, aiming to allow robots to learn complex motion planning skills in multi-goal and multi-constraint environments with a few interactions.

A goal-conditioned model-based and model-free search method with multi-timescale value function tuning is proposed in this paper. First, the authors construct a multi-goal, multi-constrained policy optimization approach that fuses model-based policy optimization with goal-conditioned, model-free learning. Soft constraints on states and controls are applied to ensure fast and stable policy iteration. Second, an uncertainty-aware multi-timescale value function learning method is proposed, which constructs a multi-timescale value function network and adaptively chooses the value function planning timescales according to the value prediction uncertainty. It implicitly reduces the value representation complexity and improves the generalization performance of the policy.

The algorithm enables physical robots to learn generalized skills in real-world environments through a handful of trials. The simulation and experimental results show that the algorithm outperforms other relevant model-based and model-free RL algorithms.

This paper combines goal-conditioned RL and the model predictive path integral method into a unified model-based policy search framework, which improves the learning efficiency and policy optimality of motor skill learning in multi-goal and multi-constrained environments. An uncertainty-aware multi-timescale value function learning and selection method is proposed to overcome long horizon problems, improve optimal policy resolution and therefore enhance the generalization ability of goal-conditioned RL.