As artificial intelligence (AI) advances at an unprecedented pace, it is increasingly vital for governments, businesses, and individuals to understand its far-reaching implications on society. Stanford Institute for Human-Centered Artificial Intelligence (HAI)’s 2023 AI Index Report provides a comprehensive analysis of the current state of AI, examining various aspects such as research contributions, model advancements, the surge in AI misuse, private investment patterns, job opportunities, industry adoption, and other trends. Here are 10 important takeaways from the report.

1. China continues to lead in AI research, consistently outpacing other countries in journal, conference, and repository publications. Although the United States retains an edge in AI conference and repository citations, this advantage is gradually diminishing. American institutions continue to make significant contributions, accounting for over half (54 percent in 2022) of the world’s large language and multimodal models.

2. Large language models are getting bigger and more expensive. GPT-2 was released in 2019 with 1.5 billion parameters and cost an estimated $50,000 to train. PaLM, launched in 2022, had 540 billion parameters and cost an estimated $8 million to train. PaLM was around 360 times larger than GPT-2 and cost 160 times more to train.
4. In 2022, generative AI made a resounding entrance into public awareness with the advent of groundbreaking text-to-image models such as DALL-E 2 and Stable Diffusion, text-to-video solutions such as Make-A-Video, and advanced chatbots such as ChatGPT. The report underscores that despite their rapid progress, these systems still grapple with the propensity to generate hallucinations and exhibit overconfidence in their outputs, leading to incoherent or erroneous responses. This raises concerns regarding their reliability in crucial applications. Nevertheless, the remarkable acceleration of advancements continues to defy expectations, bringing forth capabilities that many had previously deemed years away from being introduced.

4. The prevalence of AI misuse has markedly increased, with a notable surge in related incidents. A striking example from 2022 involved a highly convincing deepfake video that depicted Ukrainian President Volodymyr Zelenskyy seemingly admitting defeat. This upswing in occurrences underscores generative technologies’ potential to be exploited for purposes such as propaganda, disinformation, and various manipulative tactics.

5. The US leads the world in AI private investment. During 2022, more than $47.4 billion was invested in AI-related products in the US, roughly 3.5 times the amount invested in AI in China ($13.4 billion). Sectors receiving the most investment included medical and healthcare ($6.1 billion); data management, processing, and cloud ($5.9 billion); and fintech ($5.5 billion).

6. AI job opportunities are widespread in the US, with the leading three states being California (with 142,154 job postings), Texas (66,624 openings), and New York (43,899 postings).

7. Companies are increasingly integrating AI capabilities into a diverse range of business functions and departments. Among the various industries, high tech/telecom, financial services, and legal and professional services exhibited the highest adoption rates of robotic process automation at 48 percent, 47 percent, and 46 percent, respectively. Across all sectors, the most prevalent AI technologies included robotic process automation (39 percent), computer vision (34 percent), natural language text understanding (33 percent), and virtual agents (33 percent).

8. More than 27 states require that high schools offer a computer science course. The states with the most high schools teaching computer science are Maryland (98 percent), South Carolina (93 percent), and Arkansas (92 percent). This number will likely increase in the future thanks to former Arkansas Gov. Asa Hutchinson’s initiative with the National Governors Association (NGA), which secured commitments from all the nation’s governors.

9. State policymakers are also passing a steady number of laws that mention AI. Maryland has passed seven bills, followed by California (six), Massachusetts (five), and Washington (five).

10. A public opinion poll conducted in 2022 revealed that among Americans who expressed more apprehension than excitement about AI, the primary concerns include loss of human jobs (19 percent); surveillance, hacking, and digital privacy (16 percent); and lack of human connection (12 percent). On the other hand, Americans showed less concern about potential threats to personal freedom and the difficulties stemming from inadequate oversight and regulatory measures.

These findings give tangible shape to AI’s growing influence and underscore the role that domestic research, investment, and job opportunities play in giving the United States a competitive edge. This is particularly important as it relates to implementing the R&D provisions from the CHIPS and Science Act. Policymakers must balance between addressing public concerns and promoting responsible AI development to bolster economic growth and improve Americans’ quality of life.

The post Ten Takeaways from Stanford University’s Report on the State of AI appeared first on American Enterprise Institute – AEI.