and simulating a loan. The number of RBTs in Japan is 5,655 as of June 1998 according to Yano Research Institute; usually there is one RBT at one branch.
RBT is equipped with OCR and specific devices to issue a card and a passbook in addition to conventional peripherals of 2-way video conferencing systems: camera, scanner, printer, and networking facility including ISDN (Fig. 1). Usually several RBTs are connected to one center terminal which is used by a customer service representative (to be referred to as an agent). With RBT, the customer inputs necessary information through a paper form (read by OCR) and the touch screen, and gets a loan card and a passbook in less than an hour if the application passes. The agent remotely supervises this process and gives additional information and assistance to the customer from the center terminal basically on request.
RBT has two potentially conflicting functions in the awareness perspective -- whether visual and audio awareness of the agent are preferred by the customer or not. The first function is an application for a card loan, where the customer does not like to face the agent and likes to feel a sense of privacy as much as possible -- the least awareness mode for the customer of the agent. The second function is consultation on the housing loan or investment, where the customer likes presence or "awareness" of the agent to the customer -- the maximum awareness mode for the customer of the agent. In the following sections, each function (mode) is explained in detail.
In the current commercial version of RBT with the card loan contract as the main function, it is important to make the customer feel a sense of privacy as much as possible. This means that the presence or awareness of the agent to the customer is not ideal. This is because the customer does not like to request for a loan card in person or most importantly to be said in person that the application for a loan card is rejected, yet it is often the case.
Thus, the current version behaves similar to the conventional ATMs (Auto Teller Machine) because of the various input devices and conventional output methods: recorded voice, images, and texts to provide the instruction. For example, when the agent indicates incorrect fields of a paper application form, which is filled in by the customer and read by the OCR, the agent marks on the center terminal screen the incorrect fields of the OCR image in a red color. The