Let’s talk about GenAI in 2024. The things happening now that are looking likely to grow bigger during the year.

Summary

• GenAI in 2024: trends and challenges of GenAI, a term for the latest generation of AI models and applications, in the upcoming year.
• AI function in enterprises: enterprises will have more mature and clear leadership for their AI function, either centralized or decentralized, and will invest more in the enterprise stack and infrastructure to enable GenAI.
• Multimodal capabilities: emergence of multimodal GenAI, which can handle different types of data such as images, voice, video, and text, and create novel and diverse outputs.
• Data and prioritization: data leaders, data pipelines, and dataOps will become more important for GenAI projects. There will be a need to prioritize the most impactful use cases across different functions and domains.

AI function in enterprises: 

On the org-side, we’re going to see more maturing of the AI function in enterprises.

• There might be clearer leadership at the C-Suite in some orgs. With a CAIO (Chief AI Officer) having ownership of the productivity impact (for internal use) and the customer impact (for external use) with wide xfn responsibility. This role cuts across BUs, finance, operations, technology, and contact centers.
• Other orgs might go down the route of wide decentralization / enablement, with the CAIO owning the enablement function and all the function / BU leaders owning their respective impact and projects.

• The enterprise stack will mature further. I expect we will start seeing more commercial packages that will help with integration into the enterprise stack (RBAC access, security, etc). I co-wrote an article on CIO.com a couple of months ago on this orchestration layer. It’ll continue to mature.
• On the infrastructure side, the chips will get more powerful, the compute more available, and this will enable further improvements. This might mean larger context windows, more multimodal capabilities, larger LLM models, lower latency (which is an issue today).

Multimodal capabilities

Multimodal became a thing in late 2023 and we are at the early stage of this cycle. We started seeing prompting a model with a photograph and asking it to tell us something about it. Or asking a model to change a picture into something else including giving it motion. We also started seeing GenAI do things with voice, video, photos, and it is all moving really fast into production grade and with higher quality.

Data and priorization

Data leaders will get even more central billing in GenAI projects in enterprises. What enterprise data is ready for use to fine tune models? What data can be used to be fed into a RAG engine? What’s the compliance work needed to create these data pipelines and the dataOps to maintain them. A lot of conversations about getting the data space in order to capture the GenAI opportunity.

There is also a healthy discussion about horizontals prioritization. There is inquiry on what should get the lion share of investments. Is it going to be big in the contact center, in HR, in the finance function, in the tech function, accelerating programmer productivity? How do we prioritize the plethora of opportunities across the enterprise?

Finally, I think we’re going to be chipping away at the hallucinations problem. There won’t be a silver bullet, but rather continued mitigation to de-risk it further and further. Everyone knows this is an issue and there’s tons of work to mitigate it.

After a year like 2023, what will GenAI in 2024 hold? What resonates here as a theme for you?

#aistrategy #genai #aileadership