Linux Survey: 82% Say Open Source is Key to a Sustainable AI Future

Listen to an AI-generated podcast explaining the report, from Google’s NotebookLM.

TLDR

• Eight out of 10 organizations believe open source AI is “critical” to the future of AI and 61% say its benefits outweigh the associated risks.
• Asked if they saw gains from their AI investments, 62% gave answers that ranged from “some” all the way to “substantial” gains.
• Despite investments in AI, 67% said there was no impact on jobs.

A Linux Foundation survey has revealed that 82% of organizations believe open source AI is “critical” to making the technology sustainable for the future.

Moreover, 61% believe that the benefits outweigh the associated risks of open source AI, which could be used by nefarious people to commit crimes, according to the nonprofit.

In the survey, 73% of organizations expect to increase their use of open source gen AI tools over the next two years, with 26% seeing a “substantial” rise in usage.

On average, 41% of an organization’s code infrastructure supporting gen AI is open source. Central to the increase in usage are open source frameworks such as TensorFlow and PyTorch for building and training gen AI models and application frameworks.

“These open source frameworks enable organizations to build, train, and deploy models at a fraction of the cost associated with proprietary tools,” according to the report, which surveyed 316 respondents from August to September 2024 in partnership with LF AI & Data and Cloud Native Computing Foundation.

Asked if they saw any gains from their AI investments, 62% said they saw “some,” “moderate,” “significant,” or “substantial” gains. The rest saw “little or no gain.”

The top 5 frameworks for model building or training are PyTorch (63%), TensorFlow (50%), CUDA (42%), vLLM (19%), TensorFlow serving (14%). For model inference: Langchain (44%), LlamaIndex (30%), Auto-GPT (17%), AutoGen (13%), Haystack (9%).

The most important factors when choosing a gen AI model? Topping the survey is accuracy or performance (60%), security (54%), cost (49%), privacy (48%), and compliance with regulations (40%). Being open source (32%) was ninth on the list.

As for gen AI’s impact on jobs, most saw no change.

Here is the rest of the report in pictures: