The effect of AI in journalism and politics
The effect of AI in journalism and politics
Source: page-034.txt
The University of South Carolina AI Index / JULY 2024 33 THE EFFECT OF AI IN JOURNALISM AND POLITICS The survey indicates a divided outlook on AI's influence on the quality of journalism, with positive influence for 46%, negative influence for 36%, and no significant impact - for 18%. Interestingly, skepticism about AI's role in journalism increases with higher education levels and age, suggesting that more educated and older respondents are more cautious about the potential negative effects of AI on journalistic quality.
Source: page-035.txt
The University of South Carolina AI Index / JULY 2024 34 However, the overall sentiment remains somewhat positive, with a larger portion of the population leaning towards the belief that AI will enhance rather than undermine journalism quality. Figure 33 Source: Summer 2024 USC AI Index. Questions about the AI tools and the quality of journalism When discussing the impact of AI on specific areas, the survey results reveal mixed opinions among the public. 12% 34% 23% 13% 18% 0% 5% 10% 15% 20% 25% 30% 35% 40% Strongly influences for the better Moderately influences for the better Moderately influences for the worse Strongly influences for the worse No significant impact T o what extent do you believe AI can influence the quality of journalism for better or for worse?
Source: page-036.txt
The University of South Carolina AI Index / JULY 2024
35
Figure 34 Source: Summer 2024 USC AI
Index. Question - To what extent AI do you
think can contribute to online
misinformation and disinformation?
The concern about AI increasing
online misinformation is
evident, with 46% believing that
AI will heighten misinformation.
Conversely, 33% think AI might
reduce misinformation, while
21% believe AI's impact will be
neutral. The trend shows that
younger respondents feel less
at risk from misinformation,
potentially indicating a gap in
awareness about the dangers of
AI-generated content. Educated
individuals, on the other hand,
show greater apprehension about the rise of misinformation due to AI.
Figure 35 Source: Summer 2024 USC
AI Index. Question - Do you believe
that AI-generated content has the
potential to influence political
views, especially in the context of
future elections?
A significant portion of
respondents believe AI-
generated content could
influence political views,
particularly in the context of
future elections.
Specifically, 61% (37%
"probably yes" and 24%
"definitely yes") think it has
this potential. Only a small
fraction, 13%, dismiss this
possibility ("definitely not"
and "probably not"). Again, notably, younger individuals feel less exposed to the influence of AI-
generated content on political views, possibly making them more vulnerable to subtle
manipulations. In contrast, those with higher education levels are more aware and concerned
about such influences.
These insights underline the complexity of public perceptions regarding AI's role in influencing
information and politics. They highlight the need for heightened awareness and education,
especially among younger demographics, to better navigate the potential pitfalls of AI in
shaping opinions and disseminating information.
3%
10%
37%
24%
26%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Definitely not Probably not Probably yes Definitely yes Neutral
Potential to influence political views
33%
46%
21%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
50%
Reduces dis &
misinformation
Increases dis &
misinformation
AI has a neutral effect
Contribution to the
misinformation and
disinformation
Source: page-037.txt
The University of South Carolina AI Index / JULY 2024 36 Also, public perceptions on the government's role in regulating AI tools like ChatGPT are notably mixed. According to the survey, 26% worry that the government might over-regulate, potentially stifling innovation and the beneficial applications of AI. The self-labeled Republicans are even more concerned than the average American (40%). Conversely, 38% of respondents are concerned that the government will not go far enough in regulating the use of AI, indicating a desire for more stringent oversight to manage potential risks and ethical considerations. Self-labeled Democrats are interested in a higher regulation in proportion of 48%. Figure 36 Source: Summer 2024 USC AI Index. Question - As chatbots like ChatGPT become more widespread, which is your greater concern of the following? 20% 40% 26% 32% 29% 36% 48% 31% 38% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Democrat Republican AVERAGE SAMPLE Government will go too far regulating their use Neither. I am not sure Government will not go far enough regulating their use
Confidence in institutions & MEDIA
Source: page-038.txt
The University of South Carolina AI Index / JULY 2024 37 CONFIDENCE IN INSTITUTIONS & MEDIA Figure 37 Source: Summer 2024 USC AI Index. Questions about the level of confidence in institutions. Results are presented in order of the average score (higher average in top) The data on institutional trust shows a clear hierarchy, with universities enjoying the highest levels of confidence from the public, followed by YouTube, news websites, and 10% 8% 10% 12% 9% 5% 9% 13% 10% 10% 11% 14% 14% 8% 12% 11% 12% 14% 16% 16% 19% 18% 21% 20% 17% 24% 22% 28% 27% 29% 34% 36% 35% 30% 44% 40% 40% 41% 37% 27% 25% 20% 27% 26% 26% 23% 17% 19% 17% 20% 19% 15% 33% 28% 33% 20% 18% 16% 17% 21% 9% 13% 9% 9% 10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% TikTok X or Twitter The Republican Party Facebook The Government Major Corporations The press The Democratic Party Business in general Television news News websites YouTube Universities Level of confidence in institutions A great deal Quite a lot A moderate amount Not very much No confidence at all
Source: page-039.txt
The University of South Carolina AI Index / JULY 2024 38 television news. Conversely, TikTok is at the bottom of the trust scale, indicating significant skepticism about this platform. Among social media channels, YouTube stands out as the most trusted, reflecting its perceived reliability and influence. The principal component analysis (PCA) in the second graph further elucidates the public's trust patterns. It reveals distinct clusters of trust in various institutions. This analysis helps understanding the underlying patterns of trust among different demographic groups. For instance, there are clear divisions between those who trust traditional institutions like universities, TV news, and government and those who place their trust in newer, digital platforms like X, TikTok or YouTube. The PCA plot provides a visual representation of these trust dynamics, highlighting the diverse trust profiles within the surveyed population. Figure 38 A Principal Component Analysis (PCA) of survey responses concerning trust in various U.S. institutions. Each point represents an individual's trust levels, with the x-axis (PC1) capturing the most variance and the y-axis (PC2) showing the second highest variance. Clusters indicate groups with similar trust profiles, while distances between points reflect differences in trust levels. Institutions are depicted as vectors, showing their influence on the respondents' positions in the PCA space. The length and direction of each vector indicate the strength and nature of the relationship between trust in that institution and the principal components. This visualization reveals underlying patterns and distinct trust profiles among different demographic or psychographic groups.