On balance, our respondents were overwhelmingly in favor of “Buy.” - Buy or Build Depending on the Use Case - Buy - Build 96% of respondents want to buy at least some, or in the case of a third of respondents, all of their gen AI solutions, demonstrating that at least for now, there’s safety and convenience in working with the big players. But, as of now, while fewer than 15% of respondents were interested in using solely open-source LLMs (or building their own), two-thirds were interested in experimenting around fine-tuning third party options. When asked whether they want to utilize third party LLMs (e.g. OpenAI or Google) vs. build their own, this is what our respondents had to say: - Will experiment with fine-tuning third-party LLMs + augmenting them with enterprise data - Third party LLMs are good enough - Prefer to pursue open source LLMs - Prefer to build our own enterprise LLMs It’s clear from these early results that there will need to be some combination of third-party LLMs alongside those that are internally hosted and trained, and that this will likely depend on the application. In most cases, third-party will likely make the cut. For example, product- specific help for customers would be easier with a more customizable LLM, but other applications may not require that level of effort. , CIO, BCBC North Dakota “While AI looks to change the way businesses operate, each business still has its own hurdles to address to reach success. This could be data protection, cost to implement, talent, or just determining how to apply AI to the right problem.”