As we continue to discuss building an AI infrastructure for our campus, I have a few questions for the community: A. How did you decide where to start with AI? Once you did, what were the first couple of steps? B. What are the key AI roles you've started with? 

“As we continue to discuss building an AI infrastructure for our campus,
I have a few questions for the community:

A. How did you decide where to start with AI? Once you did, what were
the first couple of steps?

1) LLM Training:
  If you plan to build an AI infrastructure for your own LLM training
goal, you need to prepare for the followings.

-Real Estate/Data center building with higher maximum load weight for
cooling systems.
-Power Supply Capacity (more than conventional Data Centers)  i.e)
30MW/Building
(w/ Permission from Local Gov.)
-New liquid cooling system for GPU clusters
-Procurement plan of GPUs depended on your LLM size
(I believe the demand is very high in the market.)

Otherwise, you can use the cloud service for AI Modeling.

2) AI Inference:
  If you focus on the AI inference after building AI with Big tech's
cloud service, you do not need the LLM Training scale data center as above.

B. What are the key AI roles you've started with? ”

  The AI role would be one of the process for achieving your business
goals.
The key AI roles would be "Efficiency", "New insights", and "Cost saving".

1) Education of leadership 
2) Ideation session with leadership 
3) Impact vs Effort analysis by IT and leadership
4) Prioritization by leadership
5) Create pilots
6) Create objectives, KPIs, ways of collecting lessons learned, etc.
7) Begin pilots

Note: Many of the tools we use may have AI features (EX: resume screening in application system) that can be enabled.  Those are a great place to start your list of pilots.