Stanford AI Index shows more AI models open-source now, 52% people nervous about AI

But despite this shift, closed-source models are still outperforming open-source ones on 10 selected benchmarks.

FILE PHOTO: The 500-page report lists out various findings that reveal how AI models are moving towards open-source and which countries are the most anxious about advancements in the field. | Photo Credit: Reuters

Stanford has released this year’s AI Index tracking the major trends from the industry. The 500-page report lists out various findings that reveal how foundational AI models are moving towards open-source, how AI tools are catching up with human abilities and which countries are the most anxious about advancements in the field.

In the past year, AI firms have released 149 AI large language models - more than twice the number of models released in 2022, of which most or 65.7% were open-source as compared to 44.4% in 2022 and 33.3% in 2021. But despite this shift, closed-source models are still outperforming open-source ones on 10 selected benchmarks.

The report also stated that while AI has surpassed human-level performance on several benchmark tasks like visual reasoning, image classification and comprehension, it was still behind in more complex tasks like planning, visual commonsense reasoning and competition-level mathematics. 

AI has also succeeded with helping workers become more productive and produce better quality of work, according to the report. More companies are adopting AI in some way or the other in their business - to personalise content, acquire new clients and use AI to automate contact centers. In surveys, 55% of organisations said they were using AI last year, up from 50% in 2022 and 20% in 2017.

(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)

Additionally, despite a fall in private investments in the sector in 2023, the funding in Generative AI grew significantly to reach $25.2 billion with prominent players like OpenAI, Anthropic and Hugging Face leveraging the most of this. 

Respondents also showed a sense of unease around the proliferation of AI with younger generations anticipating more impact. In the Gen Z generation, 66% believed AI will significantly affect the current jobs compared to 46% of baby boomers. Globally, more than a third expected AI to replace them in their job roles. 

When asked whether AI products and services made them nervous, the maximum numbers showed up among Australians (69%), British (65%) and U.S. and Canada (63% for both). India ranked fourth with 58%.