Skip to content

AI Index 2025

Source: "The 2025 AI Index" report from Stanford's Institute for Human-Centered Artificial Intelligence
  • Date of Report: April 2025 (reflecting patterns and trends from 2024)
  • Overview: The 2025 AI Index, a comprehensive 456-page report from Stanford's Institute for Human-Centered Artificial Intelligence (HAI), provides a detailed, longitudinal analysis of AI's progression in 2024. The report signals a critical "inflection point," moving AI "from the realm of possibility to the realm of what's real." As co-directors Yolanda Gil and Raymond Perau conclude, "AI is no longer just a story of what's possible it's a story of what's happening now and how we are collectively shaping the future of humanity."

Main Themes and Key Takeaways:

Continuous Improvement in AI Performance:

A fundamental, albeit unsurprising, finding is that "AI performance on demanding benchmarks continues to improve." This signifies ongoing advancements in AI capabilities across various applications.

Increasing AI Embedding in Everyday Life:

AI is no longer a niche technology but is "increasingly embedded in everyday life." Beyond generative AI, the report highlights its growing presence in sectors such as transportation (e.g., Waymo) and medical devices, demonstrating a shift "from opportunity to action."

Record Business Investment and Usage Driven by Productivity Impacts:

"Business is all in on AI fueling record investment and usage as research continues to show strong productivity impacts." Investment: In 2024, "US private AI investment grew to 109 billion nearly 12 times China's 9.3 billion and 24 times the UK's 4.5 billion," positioning American enterprises at the vanguard of AI investment. Usage: Business usage is rapidly accelerating, with 78% of organisations using AI in 2024, a significant increase from 55% in 2023.

China Closing the Performance Gap with the US:

While the US maintains its lead in developing top AI models, "China is closing the performance gap." This trend, particularly evident in late 2024 and early 2025 with the launch of Deep Seek's models, represents a significant geopolitical development with potential "downstream effects."

Uneven Evolution of the Responsible AI Ecosystem:

Following a surge of interest in AI risks and concerns in 2023 (post-ChatGPT), 2024 saw a decline in focus on responsible AI, particularly in the US. The report suggests a shift towards "an out-and-out battle for AI supremacy," which, in some policy discourses, "wars with the trend right before it the rise of China." Despite a "dramatic increase in 2024" of "problematic AI" or harm-related incidents, the "raw state these numbers just aren't as big as I think people thought they were going to be," potentially leading to an underestimation of future challenges. For example, concerns about AI's impact on the US presidential elections proved largely unfounded.

Rising Global AI Optimism, Concentrated in the Developing World:

Overall, "global AI optimism is rising," but this sentiment is disproportionately high in developing nations, particularly in Asia. High Optimism: In China (83%), Indonesia (80%), and Thailand (77%), significant majorities view AI products and services as "more beneficial than harmful." Lower Optimism (but improving): In contrast, only 39% of the US population holds this view, though this marks a 4% increase. Other developed economies like Germany, France, and Canada also saw positive shifts, but the "differential between older western economies and these Southeast Asian economies is extremely notable."

AI Becoming More Efficient, Affordable, and Accessible:

AI models became "cheaper to use" and "smaller models got a lot better over the course of 2024." This dual trend has led to a significant expansion of "categories of use cases that were previously inaccessible." The cost curve for AI technologies has declined "way faster than on previous curves," making AI more accessible to a wider range of users and applications.

Rise of More Useful AI Agents:

The report highlights the "rise of more useful agents," noting a "pretty dramatic increase in the utility of agents even starting last year." This trend is expected to be a major story in future AI index reports, signifying a shift from the initial "gap between what people were talking about and what was actually possible" to tangible utility, particularly following the introduction of reasoning models.

The report concludes by emphasising key trends that are "shaping the landscape and environment in which we all operate":

  1. Corporate adoption approaching 100%.
  2. Agents becoming increasingly performant.
  3. China rapidly reaching parity with the US in AI model development.

FAQ

  • What is the Stanford AI Index and what is its main purpose? The Stanford AI Index is an annual, comprehensive report published by Stanford's Institute for Human-Centered Artificial Intelligence (HAI) since 2022. Its primary purpose is to provide a detailed, longitudinal analysis of AI's progression, trends, and patterns over the course of a year, offering insights into how the industry has evolved and where it is headed. The 2025 report, for instance, focuses on 2024 data, providing a deep dive into the shifts and developments during that period.
  • What is the overarching sentiment or "inflection point" highlighted in the 2025 AI Index? The report highlights a clear inflection point: AI is no longer merely a concept of "what's possible" but has firmly entered the realm of "what's happening now." This indicates a shift from theoretical potential to practical application and widespread integration into everyday life and various sectors, collectively shaping the future of humanity.
  • How has AI adoption and investment changed in the business world, particularly in the US? Businesses have gone "all-in" on AI, leading to record investment and usage, with research consistently showing strong productivity impacts. In 2024, US private AI investment reached an impressive $109 billion, significantly outpacing China ($9.3 billion) and the UK ($4.5 billion). Broader business usage also accelerated, with 78% of organisations using AI in 2024, up from 55% in 2023.
  • What is the significant trend concerning the US and China in AI development? While the US still holds a lead in producing top AI models, China is rapidly closing the performance gap. This trend became particularly evident in late 2024 and early 2025 with the launch of Deep Seek's models, especially their reasoning model and mobile app, marking it as one of the most significant geopolitical trends in AI.
  • What is the state of the "responsible AI" ecosystem? Interest in responsible AI and concerns around its risks, which surged in 2023 after ChatGPT, appeared to decline across 2024, particularly in the US. While there was a dramatic increase in "problematic AI" incidents, these numbers were not as high as anticipated, leading to a potential underestimation of future challenges. The focus shifted towards a battle for AI supremacy, sometimes at odds with responsible AI policy discourses.
  • How does global optimism towards AI vary across different regions? Global AI optimism is on the rise, but it is heavily concentrated in the developing world, particularly in Asia. Countries like China (83%), Indonesia (80%), and Thailand (77%) show significant majorities viewing AI products and services as more beneficial than harmful. In contrast, only 39% of people in the United States hold this view, despite a 4% increase in positive sentiment. Other developed economies like Germany, France, and Canada also saw positive shifts, but the significant differential with Southeast Asian economies is notable.
  • How has the accessibility and affordability of AI evolved? AI has become more efficient, affordable, and accessible. Smaller models significantly improved in 2024, and their usage became cheaper. This dual trend has opened up previously inaccessible categories of use cases, expanding the landscape of available AI applications. Despite the high costs associated with training new models and inference for reasoning models, the overall cost curve for these technologies has come down much faster than previous technological advancements.
  • What emerging AI development started gaining significant utility in 2024 and is expected to be a major focus in the coming years? The development of more useful AI agents started showing a dramatic increase in utility in 2024. Although the gap between public perception and actual capability was immense for a part of the year, the introduction of reasoning models significantly shifted this. This trend is expected to accelerate and become a dominant story, particularly in the 2026 AI Index report looking back on 2025.

Detailed Timeline

Pre-2022

  • Establishment of the AI Index: Stanford University's Institute for Human-Centered Artificial Intelligence (HAI) begins publishing its annual AI Index report.

2022

  • Inception of Annual AI Index: The first annual AI Index report from Stanford's HAI is released.

2023

  • ChatGPT's Impact & Initial AI Risk Focus: Following the launch of ChatGPT, there's a significant surge in interest regarding AI risks and concerns. Business AI Adoption: Approximately 55% of organisations report using AI.

2024

  • Publication of the "2025 AI Index" Report: Stanford's HAI compiles and releases its comprehensive 456-page "2025 AI Index," covering trends and data from the entirety of 2024.
  • AI Performance Improvement: AI models demonstrate continuous improvement on demanding benchmarks.
  • Increased AI Embedding in Daily Life: AI becomes more deeply integrated into everyday applications, including transportation (e.g., Waymo becoming widely available) and medical devices.
  • Record Business Investment & Usage in AI (US Leads):US private AI investment reaches $109 billion, significantly outstripping China ($9.3 billion) and the UK ($4.5 billion).
  • Business usage of AI accelerates, with 78% of organisations reporting AI adoption (up from 55% in 2023).
  • China Closes AI Performance Gap: While the US maintains its lead in producing top AI models, China makes significant progress in closing the performance gap, particularly evident towards the end of the year.
  • Uneven Evolution of Responsible AI: Interest and policy efforts surrounding responsible AI decline, particularly in the US, partly due to the distraction of the US presidential election. A shift occurs from focusing on responsible AI to a "battle for AI supremacy."
  • Rise in Global AI Optimism (Concentrated in Developing World):Optimism about AI increases globally, with a high concentration in Asian countries (China: 83% see AI as beneficial; Indonesia: 80%; Thailand: 77%).
  • In the US, only 39% see AI as more beneficial than harmful, though this is a 4% increase from the previous period.
  • AI Becomes More Efficient, Affordable, and Accessible: Smaller AI models improve significantly, and the cost of using AI models decreases, expanding the range of accessible use cases.
  • Increase in "Problematic AI" Incidents: A dramatic increase in AI harm-related incidents is observed, though the raw numbers are lower than many people anticipated.
  • Lack of AI Impact on US Presidential Election: Contrary to widespread fears, AI has "squat nothing" impact on the US presidential elections.
  • Emergence of Useful AI Agents: The utility and performance of AI agents begin to dramatically increase, particularly with the introduction of "reasoning models."

End of 2024 / Beginning of 2025:

  • Deep Seek Models Launch: Deep Seek launches its models, particularly their reasoning model and accompanying mobile app, significantly raising awareness of China's progress in AI. April 2025:

  • AI Daily Brief Discusses "2025 AI Index": The "AI daily brief" podcast/show reviews and discusses the "2025 AI Index" report, highlighting its key findings and trends from 2024. Looking Ahead (2025 - 2026):

  • Continued Dominance of AI Agent Story: The rise of more useful AI agents is predicted to be the "big story" of 2025, likely dominating the "2026 AI Index" report.

  • Corporate AI Adoption Towards 100%: Surveys are noted to likely underestimate the true extent of corporate AI adoption, with trends suggesting it is moving towards 100%.
  • China Reaching Par with US in AI: China's rapid advancement is expected to lead to it reaching parity with the US in AI capabilities.

Cast of Characters

  • Yolanda Gil: Co-director of the "2025 AI Index" report from Stanford's Institute for Human-Centered Artificial Intelligence (HAI). She contributed to the report's framing and conclusions, emphasising AI's shift from possibility to reality.
  • Raymond Perau: Co-director of the "2025 AI Index" report from Stanford's Institute for Human-Centered Artificial Intelligence (HAI). He, alongside Yolanda Gil, concluded the introduction, highlighting AI's present impact and humanity's role in shaping its future.
  • The Stanford Researchers: The collective term for the individuals who compiled and authored the "2025 AI Index" report at Stanford University's Institute for Human-Centered Artificial Intelligence (HAI). Their work forms the basis of the entire source.
  • Swix: An individual with whom the narrator of the "AI daily brief" had a conversation, scheduled for the following week, discussing AI agents and their capabilities in 2024. (Note: Only mentioned in the context of a conversation, no further personal details provided).