The author suggests that the next frontier in generative AI includes intent mapping, domain-specific indexing, and non-textual answers. These methods could help AI understand what information is being sought and decide what information to bring to the model. The author concludes that generative AI with contextual business knowledge could potentially drive a level of innovation and growth that we have yet to witness.
Key takeaways:
- Generative AI needs to provide deeper value by marrying an organization’s local data with AI capabilities and providing company-specific and context-specific outputs.
- Several methods have been tried to make generative AI more valuable to business outcomes, including increasing prompt size, training ad hoc LLMs, retrieval-augmented generation (RAG), and function calling.
- For generative AI to deliver value by addressing non-trivial business questions, it must understand what information is being sought, and then decide what information to bring to the model. This can include intent mapping, domain-specific indexing, and non-textual answers.
- Generative AI with contextual business knowledge could allow businesses to leverage their data in previously unimaginable ways, potentially driving a level of innovation and growth that we have yet to witness.