Quality Alchemist (品質煉金術師): Leader 2030 Webinar – How Are AI Projects Different from Traditional IT Projects

2025年3月19日星期三

Leader 2030 Webinar – How Are AI Projects Different from Traditional IT Projects

The Executive Study Group (ESG) webinar named “Leader 2030 – How Are AI Projects Different from Traditional IT Projects” was held by Asia Pacific Institute for Strategy Limited (APIFS) on 19^th Mar 2025. Before the webinar, Dr. Mark Lee introduced the mission of Leader 2030 that aims to develop leadership capabilities for 2030s and build a strong network of future-focused leaders. Then he introduced the guest speaker Mr. Eric Moy (CEO, KB Quest).

In the beginning, Mr. Eric Moy briefed the trend from Large Language Models (LLM) to Large Reasoning Models (LRM). Only DeepSeek is open source in the top level of LRM.

Then he mentioned from Agentic to AGI that AGI possesses the ability to understand, learn, and apply knowledge across tasks at a level comparable to human intelligence. And then he demonstrated the latest AI Agent “Manus” as example. It showed how to compare NVDA, MRVL & TSM stock prices and wrote website to demonstrate it.

After that Mr. Eric Moy compared the definition between traditional IT project and AI project. The key differences are from rule-based decision to Agentic AI / AI making decision.

He said AI projects’ scope was difficult to define and framework was still evolving.

Different skills and team composition included data scientist, machine learning engineer and domain experts that emphasized on research and experimentation but not focused on coding and system integration.

Testing of traditional IT project and AI project were different from focus on functional testing to model accuracy. Where data were core component in AI project.

Finally, Mr. Eric Moy mentioned three dimensions to be considered as AI project completion that were accuracy, time and cost. The most important one is to establish a hard deadline! Maintenance is also considered as ongoing such as new version of GenAI launch.

Lastly, he reviewed five factors of distinctions.

He recommended AI project flow from coaching to executive for expectation management and then defined the firm deadline, after that define minimal acceptable standards and ensure executive involvement.