DAI can be exploited in knowledge acquisition for modelling the experts
( their cooperation, their conflicts), for modelling the knowledge
acquisition process and specially
the cooperation during knowledge acquisition from a group of experts.
(Similar concepts are used in process modelling, see 
).
For a survey, see [Doer91][AAAI91].
In [Dieng91a], the process of knowledge acquisition is described as the behaviour of a composite system made of several human agents (experts, knowledge engineers and users) and software agents (the knowledge acquisition tool, the final system and the software where the final system will be integrated). This notion of composite system made of heterogeneous, interacting agents can help to model the main relations between such agents (knowledge transfer, explanation, validation, assistance to problem solving...) and to analyse the cooperation underlying the process of knowledge acquisition. The notion of agent should allow to model the end-user as an agent and ease the description of a knowledge acquisition methodology involving several human agents.
In [Wern92b], the problems of design of a multi-agent system with multiple interaction levels are studied. It enables to specify the interaction points among the multiple agents involved in such a design and the interaction protocols among such agents. Such research can be extended in order to guide the construction of a knowledge acquisition tool which is based on a multi-agent architecture and involves several heterogeneous agents: human agents such as the experts, the knowledge engineers, the users, and artificial agents such as the knowledge acquisition too and the cognitive agents composing it.
The logics of intentions and of capabilities [Wern90] can be used for specifying the interactions among multiple cooperating agents that may be either human experts from whom knowledge must be acquired or cognitive agents that will represent them in a knowledge acquisition tool. The use of such a logics may enable to formalize research on knowledge acquisition from multiple experts.
The model for conflict management and resolution in the software design process proposed in [East91] can easily be extended to distributed knowledge acquisition: the author proposes techniques for comparing several viewpoints and for solving conflicts among them. In [East89] a multi-agent architecture allows to perform distributed knowledge acquisition.
In [Dieng92c], a model of cognitive agent is proposed in order to take into account knowledge acquisition from multiple experts.