People who feel strongly optimistic about AI tend to agree with one another across different questions. People who are strongly concerned about AI are less consistent.
That is the clearest pattern I found in a small exploratory analysis of six questions from the 2024 U.S. General Social Survey.
Using a Response Item Network (ResIN), I represented each possible answer as a node in a network. Responses that frequently occurred together were connected. The result suggests that strong optimism forms a compact package of beliefs, while concern and neutrality allow for a wider variety of combinations.
From survey answers to a network
A conventional survey analysis might assign each person an overall score for “AI optimism.” This is useful, but it can hide the structure of the answers.
ResIN takes a different approach. It treats every combination of a question and a response as meaningful.
For example, “not at all worried” about AI-related job losses is not classified from the wording “not at all” alone. Its meaning depends on the question: in this case, it expresses a positive or reassuring attitude toward AI. Conversely, agreeing that AI will create serious risks expresses concern.
I classified the response options into five interpretive levels: strong concern, moderate concern, neutral or uncertain, moderate optimism, and strong optimism. Responses expressing similar ideas can also be combined—for example, “agree” and “strongly agree”—when that produces a clearer and theoretically defensible category.
What the network shows
The six questions generate 15 possible connections between distinct items at each attitude level.
Among the strongly optimistic responses, all 15 of 15 possible connections remained in the network. Someone expressing strong optimism about one aspect of AI was therefore also relatively likely to express strong optimism about the others.
Among the strongly concerned responses, only 9 of 15 possible connections remained.
This does not mean that concerned respondents were inconsistent or irrational. It means their concerns were more differentiated. Someone might worry about AI replacing jobs, for instance, without expressing the same degree of concern about every other technological consequence.
The middle of the network reinforces this interpretation. Neutral responses were connected 2.6 times more strongly to concern than to optimism.
Neutrality therefore did not appear to be an evenly balanced midpoint. It looked more like cautious uncertainty, overlapping with concern while still allowing many mixed combinations of beliefs.
Why this is interesting
Public debates often describe attitudes toward AI as a single spectrum running from enthusiasm to fear. This network suggests that the two sides may not be mirror images.
Strong optimism appears to function as a coherent worldview: positive expectations in one area tend to accompany positive expectations elsewhere.
Concern is more fragmented. Different people may be concerned about different things, for different reasons. Between neutrality and strong concern, there is no sharp boundary—there is a larger space of mixed beliefs.
That distinction matters for communication. A single reassuring argument may reinforce an already coherent optimistic outlook, but it is unlikely to address every form of concern. Understanding skepticism may require asking what specifically someone is worried about rather than treating all negative attitudes as one position.
A quick exploratory study
This is an AI-assisted exploratory analysis, not a full peer-reviewed study. The classifications, network settings and interpretation should be tested through robustness checks and replicated with other datasets.
Still, it illustrates what ResIN can add to survey research. Instead of reducing people to one score, it helps us see how specific answers fit together—and where apparently simple attitude categories contain very different belief structures.
Dataset: 2024 U.S. General Social Survey