Human-Centered Artificial Intelligence (HAI) is the AI research unit of Stanford University that has released the annual Artificial Intelligence Index Report 2024, the 7th edition of this report. The report compiles data on the development of AI from various sources, including the impacts that have occurred.
Key points in this year’s report are varied and include:
AI surpasses human abilities but is not perfect. In testing 9 HAI categories, AI has shown superiority over human standards in various aspects (some for a long time now), such as image recognition, reading comprehension, image reasoning, and English language interpretation. However, in some areas, AI still performs poorly compared to humans, especially tasks that require complex contextual understanding, such as reasoning based on common sense or solving challenging math problems.
AI research has seen a significant increase. The research in AI has seen a substantial increase in the past few years. In 2020, there were approximately 88,000 topics, while in 2022, there were up to 240,000 topics. Notably, the standout Machine Learning models from the past year included 51 models from the private sector, 21 from collaborative efforts between educational and private institutions, and 15 models from the education sector. The majority of these models originate from the United States, followed by China in the second position.
AI remains costly, but investors are willing to pay. The report also discusses the costs of leading AI models in the market. The estimated figures are exceptionally high, with the training cost of Gemini Ultra around $191.4 million, and GPT-4 approximately $78.3 million. These costs vary depending on the number of processors used for processing. The total investment amount in AI companies stands at $1.89 trillion, a 31% decrease from 2022. However, compared to 2013 or a decade ago, this figure has increased by a factor of 13. Investment from venture capital into AI startups in 2023 is at $9.599 trillion, slightly down by 7% from 2022. When focusing on Generative AI, the investment in 2023 has significantly increased by 5-6 times from 2022, amounting to $2.5 trillion.
AI may pose a threat to human jobs, but it greatly benefits the scientific front. A global survey conducted by Ipsos on how people view the impacts of AI revealed increased awareness, with 66% (up from 60%) saying that AI will impact their lives in 3-5 years, while 52% are concerned about AI’s abilities to impact various aspects. Despite concerns about AI, advanced scientific research has greatly benefited from AI. In 2023, several research projects utilized AI to achieve quicker results, such as AlphaDev for high-speed algorithm sorting, FlexiCubes for 3D modeling, GraphCast for weather forecasting, GNoME for discovering new materials, and many more. AI has accelerated the processing speed in these research endeavors.
TLDR: The annual Artificial Intelligence Index Report 2024 by Human-Centered Artificial Intelligence at Stanford University showcases how AI surpasses human abilities in various aspects, while also experiencing significant growth in research and investment. Concerns about job displacement are balanced by the benefits AI brings to advanced scientific research.
Leave a Comment