AI Among Hispanics: What the Data Shows, and What It Doesn’t Yet Understand

By Maria Lucia Parra

AI adoption has moved quickly across the U.S., and Hispanics are part of that momentum. But looking at adoption alone misses the more important question: how well are these tools actually understanding the people using them?

To answer that, I reviewed multiple waves of the MRx Pros Omnibus Study conducted by MFour Data Research (n=2,000 adults per wave, P18+), focusing specifically on Hispanic respondents and comparing them to the total market.

Within this sample, Hispanics represent a diverse group even at a directional level: 14% are first-generation immigrants and 86% are U.S.-born. Language usage further reflects this diversity, with 38% bilingual, 34% English-dominant, and 28% Spanish-dominant. Even within a standard omnibus, this already demonstrates that Hispanics are not a single profile but a combination of different experiences, identities, and ways of navigating culture.

Across waves, the results are highly consistent, which allows us to focus not on isolated data points but on patterns that hold over time.

Three key findings stand out.

1. Hispanics are not behind in AI—they are already using it at the same level as the general market.

Across all waves, roughly 60–65% of Hispanics report using AI tools like ChatGPT, Copilot, or Gemini often or occasionally. This aligns closely with the total market, where adoption falls within the same range, indicating no meaningful difference in access or participation.

This is important because it challenges a common assumption in multicultural research that Hispanic audiences are still in earlier stages of technology adoption. In this case, that is not supported by the data. Hispanics are already active users of AI, and that usage is stable over time. From a research perspective, this shifts the focus away from adoption and toward experience, expectations, and interpretation.

2. Hispanics are not just using AI to search—they are using it to do.

While information and research remain the top use cases for both Hispanics and the general market, Hispanics show a consistent pattern of slightly under-indexing in pure information use and over-indexing in more applied use cases, such as work tasks, learning, and writing support, particularly in earlier waves.

Over time, both Hispanics and the general market show a shift toward more exploratory behaviors, including curiosity-driven and creative use. However, among Hispanics, the balance between functional and exploratory use suggests that AI is already embedded in how tasks are completed, not just how information is accessed.

This distinction matters because it reflects a deeper level of integration. AI is not simply replacing search; it is supporting output, decision-making, and everyday problem-solving. For researchers, this means AI is already influencing how people process information and act on it, not just how they find it.

3. AI performs well functionally, but it is more likely to misrepresent Hispanics than the general market.

Across waves, approximately 33%–36% of Hispanics report that AI has made assumptions about them. In the total market, this number is consistently lower, typically in the high 20s to low 30s. At the same time, Hispanics report similar levels of alignment in tone and communication style, with roughly 85%–90% saying AI feels in tune at least sometimes.

Taken together, these results point to a clear and consistent tension. AI is working in terms of usability and interaction, but it is not always accurately representing the people using it. This gap does not narrow over time, suggesting it is not a temporary issue related to early adoption but rather a structural limitation.

This also helps explain behavioral differences. Hispanics are slightly more likely to verify AI outputs by checking other sources, asking others, or cross-referencing information. This is not a rejection of AI, but an adjustment to a tool that is useful but not fully reliable in its reflection of identity and context.

What this means for understanding Hispanics—and for research

These findings point to a gap that becomes more significant when we move from usage to interpretation.

AI tools are performing well at a functional level. They are accessible, efficient, and increasingly integrated into daily behavior. However, the data shows that they are still more likely to make assumptions when interacting with Hispanics than with the general market.

That difference matters because of the nature of the Hispanic population itself. Hispanics are not a homogeneous segment. Differences in country of origin, generation, language dominance, identity, and cultural expression shape behavior in ways that are not captured by surface-level variables. Even within this sample, the distribution of U.S.-born, first-generation, bilingual, English-dominant, and Spanish-dominant respondents reflects that complexity.

This is precisely why research frameworks have evolved beyond language and country of origin alone. Variables such as cultural affinity exist because they better capture how culture is actually lived and how it influences decisions. At this stage, AI is not consistently capturing that level of nuance.

This has direct implications for research. AI is increasingly being used to summarize findings, generate insights, and support analysis. However, if the same systems that are more likely to make assumptions about Hispanic users are also used to interpret them, there is a clear risk that outputs will reflect simplified or incomplete views of the audience.

The issue is not that AI is wrong. It is that it can be directionally correct while missing critical context. That distinction is essential in multicultural research.

AI Is Scaling Fast. Understanding Hispanics Is Not.

These findings point to a gap that becomes more significant when we move from usage to interpretation. AI tools are performing well at a functional level. They are accessible, efficient, and increasingly integrated into daily behavior. However, the data shows that they are still more likely to make assumptions when interacting with Hispanics than with the general market.

That difference matters because of the nature of the Hispanic population itself. Hispanics are not a homogeneous segment. Differences in generation, language dominance, identity, and cultural expression shape behavior in ways that are not captured by surface-level variables. Even within this sample, the distribution of U.S.-born, first-generation, bilingual, English-dominant, and Spanish-dominant respondents reflects that complexity.

This is precisely why research frameworks have evolved beyond language and country of origin alone. Variables such as cultural affinity exist because they better capture how culture is actually lived and how it influences decisions. At this stage, AI is not consistently capturing that level of nuance.

This has direct implications for research. AI is increasingly being used to summarize findings, generate insights, and support analysis. However, if the same systems that are more likely to make assumptions about Hispanic users are also used to interpret them, there is a clear risk that outputs will reflect simplified or incomplete views of the audience.

The issue is not that AI is wrong. It is that it can be directionally correct while missing critical context. That distinction is essential in multicultural research, where nuance is not optional; it is the insight.

At Ingenium, we see this every day. Understanding Hispanics requires going beyond surface-level variables to how culture is actually lived, and how identity, context, and experience shape decisions. AI can support the process, but it cannot replace the interpretation that comes from a deep understanding of the audience. If the goal is to truly understand Hispanic consumers, not just describe them, human expertise is still essential.

AI is already part of the process. Understanding is where the real work begins.