Page 572 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

on obtaining high-quality transactional data to further train the AI for better decision-making
during price negotiations.

2. Computational Power for Scaling: As the system becomes more advanced, the need
for enhanced computational resources will grow. More powerful cloud infrastructure and data
processing capabilities will be necessary to manage the increased volume of real-time data,
complex multi-agent systems, and advanced machine learning algorithms.

3. Human Expertise in Continuous Improvement: Human input will continue to be valuable
for tuning the AI’s negotiation strategies and ensuring its adaptability. Experts in negotiation and
AI model development will need to review the outcomes and fine-tune the system periodically.
Additionally, feedback loops from real-world negotiators will ensure that the AI's performance
aligns with industry standards.

Potential Future Collaborations
1. Industry Partnerships for Data Sharing and Expansion: Hyundai Motors, Byte Dance,
KT(Korea Telecom)

2. Collaborations with Academic and Research Institutions: Ongoing research
collaborations, particularly with Korea University HI-AI Research and other institutions, will drive
the development of advanced natural language processing (NLP) technologies and machine
learning models. These collaborations will also introduce the integration of Long-Term Memory
(LTM) and Short-Term Memory (STM) in the negotiation system [1]. By applying LTM and STM,
the AI will have enhanced capabilities to retain and recall negotiation history, which will improve
its decision-making processes. LTM will store long-term strategic learnings and insights gained
from past negotiations, while STM will focus on immediate negotiation data, allowing the AI to
respond more effectively to ongoing interactions. This memory-based approach will enable
the AI to adapt strategies based on both immediate and historical contexts [4].

3 Use Case Requirements

Technical Requirements:

REQ-01: It is critical that the system be implemented as well as deployed to seamlessly integrate
with KT Commerce's existing back-end database to facilitate real-time data exchange and
ensure smooth operation across platforms.

REQ-02: It is expected that the system be implemented but may not be deployed to support
real-time communication between the AI and human negotiators, ensuring that the negotiation
progresses without delays and adapts to changes in real time.

REQ-03: It is critical that the system be implemented as well as deployed to offer an intuitive
and easy-to-navigate interface for both MDs and customers to engage in price negotiations
efficiently.

REQ-04: It is of added value that the AI be capable of dynamically adjusting its persona based
on the negotiation data, such as the negotiation partner’s profile, market conditions, and past
negotiation outcomes. This need not be implemented or deployed immediately.

536