Could you share personal insights or resources on what part of the business organization is standing up AI capability? Is it in Data, or Tech, or another part? 

All three, and with 'another part' representing 'business'. Teams closes to business opportunities, and often operating under 'constraints' are largely coming up with more 'micro-AI' capabilities that address distinct business opportunities linked to both productivity gains and innovation. We're not stopping this, but working to put good governance and oversight around this to support teams solving challenges but ensuring good-practice and design, avoiding duplication of effort and being there to partner to ensure reusability and scalability are part of the plan. 

For this comment focusing on LLM as AI capability is this is what is most talked about nowadays.

We started our journey about a year back, focusing on the lowest hanging - customer support, starting from internal use to assist our customers and then expanding to an automated interface. I don't think its just data, but rather "Content" and how content is structured to build the knowledge base, it's chunking and content organization for security and other precautions (proper guard rails)

The steps we took were 
1. Get a tech stack that could deliver LLM / RAG / Chat UX / Embedded within our UI
2. Use content from the knowledge base, make it work - control hallucinations
3. Use content and extend with the Large Action Model to gather data during the conversation. Some success with multi-prompts / multi-model. LAM will help you merge a conversation with data in a meaningful way. - control hallucinations
4. What I feel is most important is User Adoption. We are now spending most of our time focusing on engaging users with the chat / UX interface as there was a drop post the initial adoption jump. I believe this is something that even chatgpt saw in it initial days of adoption.
5. Iterate, Iterate, Iterate

So I would put the sequence as Tech > Content > UX > User Adoption > Reporting

Hope it helps

Who is, and who should be are an interesting set of questions.  When you think about the intersectionality of human and machine (AI) and what that means for the future of work, then the HR department should definitely be a big part of this conversation.  But it seems they're not.  It seems it sits often in the Automation/IT/Data space.  Come on HR practitioners - it's time to take some of the reins here!

In our organization (Atalan Tech), our AI capability is core to our product and therefore it is developed as a collaboration between Product, Data, Science/Research and Engineering . We believe this cross-functional approach to be fairly standard and necessary to have a successfully leverage AI in our products and services. We are a health tech start-up focused on clinician wellbeing and retention.    

Multiple organizations are driving adoption of AI capabilities. For example, lines of business are pursuing AI solutions to their problems in collaboration with Legal, Technology, Privacy, and Data Governance, e.g. x-ray classification, while Technology is leading adoption of AI-enabled platforms and shared services, such as enhanced collaboration and communication capabilities. Governance and oversight of AI adoption is provided by a cross-functional committee with representatives from all major risk owners.