Marketing Roundup: 10 Highlights From Stanford’s Annual AI Index Report 

Highlights From Stanfords Annual AI Index Report

Stanford University has released its seventh Artificial Intelligence (AI) Index report. As the most comprehensive edition to date, it makes for a valuable resource for any marketing executive or professional learning to navigate the complex AI landscape. 

Loaded with carefully reviewed, widely collected data, the 500-page report provides ongoing, objective snapshots of AI development across several key areas. From technical advancements and private investments to policy initiatives and public sentiment, it captures the full scope and impact of this transformative technology. 

Some of these trends and insights can help shape the marketing agenda today and in the future. 

AI Exceeds Human Ability To Perform Various Tasks 

AI systems have outperformed humans on several intellectual task categories. These tasks include image classification, visual reasoning, basic reading comprehension, and English understanding. Still, despite making a ton of progress in 2023, AI trails behind on more complex cognitive tasks like visual commonsense reasoning and competition-level mathematics. 

AI Makes Workers More Productive and Enhances Work Quality 

The report points to data suggesting AI’s potential to boost productivity. Numerous studies on AI’s impact on labor show how current AI capabilities enable workers to complete tasks faster and more efficiently, increasing their output quality. In addition, these studies demonstrate how AI technologies may empower less skilled workers to perform tasks previously reserved for more skilled workers.  

Meanwhile, other studies caution against using AI without proper oversight, which can contribute to diminished performance. 

AI Helps Businesses Improve Their Bottom Line 

A new McKinsey survey reveals that 42% of organizations (up by 10% from last year) claim that implementing AI, including generative AI, has reduced their expenses. Cost savings specifically apply to manufacturing, service operations, and risk. Another 59% report that their revenue is increasing, which benefits manufacturing, marketing and sales, and risk. 

Industry Continues To Dominate Frontier AI Research 

Since 2014, industry has replaced academia as the leader in producing powerful frontier models. In 2023, it contributed 51 machine learning models compared to academia’s 15. Interestingly, last year saw 21 notable models emerge from industry-academia collaborations, a new high. 

Industry’s dominance can be attributed to greater access to substantial amounts of data, computing power, and financial resources — key components to building cutting-edge machine learning models. 

Model Training Costs Reach Unprecedented Levels 

New AI Index estimates indicate that training costs of state-of-the-art AI models are rising. In 2023, the training cost for OpenAI’s GPT-4 is estimated to be around $78 million. For Google’s Gemini Ultra, the training cost is approximately $191 million. In contrast, the original Transformer model, which laid the foundation for virtually every modern large language model (LLM), cost around $900 to train. 

Generative AI Investment Skyrockets 

AI-related private investment has plummeted for two years in a row. On the bright side: funding for generative AI surged to $25.2 billion, a nearly ninefold increase compared to the investment of 2022. Led by the likes of OpenAI, Anthropic, Hugging Face, and Inflection, players in the generative AI space accounted for a quarter of AI-related private investment. 

U.S. Remains the Leading Source of Top AI Models 

Delving into the country of origin of notable models, the AI Index research team has mapped the geopolitical landscape of AI development. With 61 notable machine learning models, the U.S. outranks other major geographic regions — a record it has set since 2003.  

The European Union and United Kingdom collectively secured second place with 25, surpassing China, which has 15, for the first time since 2019.    

Current AI reporting practices lack robust and standardized evaluations 

New analysis from the AI index suggests a lack of standardized benchmark reporting for evaluating AI capabilities responsibly. Leading AI model developers, specifically OpenAI, Meta, Anthropic, Google, and Mistral AI, rely on different benchmarks to test AI capabilities. This practice complicates systematically comparing top AI models’ risks and limitations. 

AI sparks increasing legislative interest 

More laws mentioning AI were enacted last year. For instance, Belgium passed five AI-related bills into law, followed by France, South Korea, and the United Kingdom with three. Since 2016, the U.S. has established 23 AI-related laws, the most any country has set.  

These laws cover a wide range of AI applications, from data privacy to autonomous vehicles, reflecting the growing legislative interest in AI. 

People across the globe become more aware of and nervous about AI’s potential impact 

A 2023 survey from Ipsos reveals a significant shift in public perception. 66% of respondents anticipate a dramatic impact of AI on their lives in the next three to five years, a 6% increase from the previous year.  

More importantly, 52% express concern about AI products and services, which aligns with Pew data indicating a growing unease among Americans about AI. This global awareness and concern underscore the urgency and importance of understanding and managing AI’s impact. 

These are some of the macro trends that can affect the marketing agenda in 2023. For a more comprehensive study of the evolving AI ecosystem, read the entire Artificial Intelligence Index Report 2024


Join Now and Elevate Your AI Game

Join the 500,000 others who receive our Scienz AI Spark Newsletter delivered to your mailbox to get the latest AI news, expert interviews, tools, reviews, and in-depth articles. You’ll also be the first to know when we launch new features and opportunities.