The famous Stanford University has just published its AI Index Report. A sum of more than 500 pages which has become, de facto, the world reference on the state of development of Artificial Intelligence and industry.
Summary in 10 key points:
1- AI does not yet surpass humans
Artificial intelligence (AI) is effective in some areas, but not yet in all. AI excels at image analysis, visual understanding and language. Future AI challenges include advanced mathematics, logical reasoning, and planning.
2- The private sector dominates research
Private industry continues to dominate cutting-edge research in artificial intelligence (AI). In 2023, private industry produced 51 machine learning models, while academia contributed only 15.
21 models were the subject of joint development between companies and universities.
3- Theincreasing inflation of learning models
Avant-garde models are becoming more and more expensive. According to AI Index estimates, the costs of training cutting-edge AI models have reached unprecedented levels. For example, OpenAI’s GPT-4 learning model took approximately $78 million while Google’s Gemini Ultra cost 191 million.
4– The United States is in the lead ahead of Europe and China
The United States has 61 LLM models compared to 21 in Europe and 15 in China.
5– No standards for responsible AI
There are not yet sufficiently robust and standardized models for responsible AI development.
Open AI, Google and Anthropic test their models on different criteria.
6– Explosion of investments in generative AI
Investments in generative AI have increased fivefold since 2022 to reach $25.2 billion.
The main players in generative AI are OpenAI, Anthropic, Hugging Face and Inflection.
7- Notable improvements in productivity and quality
In the world of work, all studies conclude that there has been a notable increase in the quality and productivity of Artificial Intelligence.
However, some studies warn of a decline in performance without appropriate supervision.
8- An acceleration of scientific progress
Since 2022, driven by dedicated models such as AlphaDev or GNoMe, new AI models have significantly accelerated scientific discovery.
9- Regulatory catch-up
The United States went from just one AI law to 25 by 2023.
Over one year, the number of regulations linked to Artificial Intelligence increased by 56.3%.
10- A growing concern among the general public
An Ipsos poll reveals that in one year, the proportion of people believing that artificial intelligence (AI) will radically affect their lives in the next three to five years has increased from 60% to 66%.
Additionally, 52% of respondents say they are nervous about AI products and services, an increase of 13 percentage points from 2022. In the United States, data from the Pew Research Center indicates that 52% of Americans say they are more worried than enthusiastic about AI, compared to 38% in 2022.
Unsplash Photo Credits by Allen Gong
