The future of generative AI lies in its application to specific domains and contexts, rather than aiming for generalized AI. While there has been hype surrounding generative AI, it’s important not to overlook immediate risks such as sustainability and bias. Generative AI systems, like ChatGPT, can be valuable tools for niche domains by providing novel ways of finding and exploring specific information. Organizations, such as Expedia, can leverage these tools to gain a competitive edge in information discovery. OpenAI is capitalizing on this opportunity by offering access to its generative AI systems, but there are also options for self-hosted large language models (LLMs) that provide privacy advantages. The future of generative AI will involve domain-specific language models and fine-tuning publicly available LLMs on specific data to develop useful information retrieval tools. Overall, the success of generative AI lies in its integration into specific contexts, where it can become a useful and embedded technology.
| Signal | Change | 10y horizon | Driving force |
|---|---|---|---|
| Generative AI shifting to niche domains | From generalized AI to specific domains | More specialized AI tools for specific domains | Need for novel ways of finding specific info |
| OpenAI’s ChatGPT plugin initiative | More organizations adopting generative AI systems | More products and interfaces backed by AI | Commercial opportunity and market advantage |
| Self-hosted LLMs | Less reliance on OpenAI for generative AI technology | Increased privacy and data control | Concerns with privacy and data management |
| Domain-specific language models emerging | Fine-tuning LLMs for specific information retrieval | More useful information retrieval tools | Improved customer support and content creation |
| Subtle shift in relationship with AI | AI becoming embedded in context | AI losing mystique and becoming useful | Realization that AI doesn’t know everything |