AI Index 2026: Experts and Public Divided on AI

The United States Leads in AI Investment

The latest report from the AI Index at Stanford, recently published, highlights striking statistics on investment in artificial intelligence. The United States stands out particularly with its 5,427 data centers, a number that continues to grow. This figure is more than ten times that of any other country, underscoring the dominant position of the United States in the field of AI. This dominance is accompanied by a continuous increase in the number of these facilities, further strengthening their technological leadership.

Global Dependence on TSMC for AI Chips

The report also underscores critical vulnerabilities in the AI hardware supply chain. A particularly striking point is the global dependence on a single company, TSMC, which manufactures nearly all of the leading AI chips. This concentration makes the supply chain extremely reliant on this unique foundry located in Taiwan, posing potential risks to the stability of the industry. The fact that a single foundry is responsible for such a significant share of global AI chip production is a major bottleneck that could have global repercussions in the event of a disruption.

Inconsistency in Perceptions of AI

The AI Index 2026 reveals another intriguing facet of AI: its current state is marked by inconsistencies. As Michelle Kim points out in her analysis of the report, AI is perceived both as an economic opportunity and a threat to employment. For instance, Google DeepMind's Gemini Deep Think model won a gold medal in mathematics but often fails to read analog clocks, illustrating the current limitations of AI. This duality in AI capabilities shows how the technology can be both advanced and limited, depending on the context of use.

A Gap Between Experts and the General Public

The report highlights a significant gap between the perceptions of AI experts and those of the general public. According to the AI Index, 73% of American experts view AI positively for employment, compared to only 23% of the public, creating a 50 percentage point gap. Similar divergences appear regarding the economy and healthcare. This perception gap may be due to differences in access to information and direct experience with AI between these two groups.

Reasons for the Divergence in Perception

This difference in perception could be explained by the varied experiences of experts and the general public with AI. A developer recently noted on X that the wonder surrounding AI is often linked to the frequency of using AI for technical tasks like coding. Current models excel in these areas, which can positively influence the perception of regular users. Experts, who frequently interact with advanced AI models, may have a more optimistic view of their capabilities.

Variable Performance of AI Models

AI models, particularly LLMs (large language models), show uneven performance. While they are effective for technical tasks, they still make errors in other contexts. This phenomenon, referred to as the "irregular frontier," reflects the variability of AI performance across different application domains. This irregularity in performance may contribute to the mixed perception of AI among the general public.

Insights from Influential Experts

Andrej Karpathy, an influential AI researcher, shared his observations on this divergence in understanding AI capabilities. He noted that advanced users, who invest up to $200 per month for the highest-performing versions of LLMs, experience spectacular improvements, which is not necessarily the case for the general public. This difference in experience between advanced and casual AI users may partly explain why perceptions diverge so significantly.

Two Realities of AI

In conclusion, AI presents two distinct realities. On one hand, it is far more advanced than many imagine, particularly in technical fields. On the other hand, it remains limited in areas that concern the public. This duality must be taken into account by those contemplating the future of AI. Policymakers and businesses must navigate these perceptions to maximize the benefits of AI while mitigating its limitations.