AI index report

Came across this series of reports produced by Stanford University.

It is an AI index, tracking the reach and implications of AI. The reports go back to 2017. 

In the 2024 report, ten 'takeaways' are distilled as to the status of AI adoption, utilisation and recommendations.

The takeaways temper the hype revolving around AI currently, reminding us that AI still has a way to go but has started to make accelerate scientific work.

The takeaways are:

- AI is able to do and beat humans at some tasks. Tasks include image classification, visual reasoning and English understanding. However AI still lacking in 'common sense' and not able to undertake complex tasks exampled by competition-level mathematics.

- Industry continues to dominate AI research, instead of academia.

- Training costs for AI models are expensive. ChatGPT-4 costs $78million and Gemini Ultra $191 million.

- The US leads as the leading source of AI. China, EU and UK also active.

- Robust and standardised evaluations of LLM responsibility lacking.

- investment in Gen AI skyrockets.

- AI helps make workers more productive and can lead to higher quality work. 

- scientific progress accelerates, facilitated by AI.

- AI regulations in the US of A increases sharply.

- Greater awareness internationally of AI's potential impact, and more people report becoming more concern rather than excitement.

Will keep an eye on these reports as they help provide a holistic picture of AI's rise, adoption, and challenges.