Source: DC Studio/Technology/Adobe Stock.
“Is it live, or is it Memorex?” That was an advertising slogan for a brand of cassette tapes in the 1980s that claimed their sound quality was as good as a live performance. Fast forward 40 years and consumers are asking a similar question about AI-generated content, which has turned our relationship with media and entertainment upside down. And now that the AI genie is out of the bottle, it looks like there is no going back.
Generative AI is capable of churning out a wide variety of content. A large share of the content revolves around consumers experimenting with their newfound ability to create fun pieces of art or prose. However, the sheen fades and real concerns emerge when those efforts are turned toward more serious endeavors.
This report presents select findings from a S&P Global Market Intelligence 451 Research’s study, which asked US consumers about their attitudes toward content produced with the use of generative AI.
Overall, respondents are split
Like most new products or tools, the adoption of generative AI hinges on the confidence users have in the output. This might not matter as much for someone who is using it for personal projects and can trial-and-error their way to a desired result. However, it becomes crucially important when generative AI is being used to create and disseminate important information.
- When asked how confident they are in the accuracy and reliability of published content produced by generative AI, just over half (53%) of respondents said they are very or somewhat confident in the results, while the balance (47%) said they are not very or not at all confident in the results.
- Likewise, when asked if content produced by generative AI is more or less accurate than content produced by human writers, 38% said generative AI produces more accurate content, while 37% said humans are more accurate. The last 25% said there is little to no difference in accuracy.
This data shows that roughly half of society believes in the accuracy of generative AI and that it may even be better than humans, while the other half are more pessimistic. And there are legitimate concerns. These AI models and their output are only as good as the data they’re trained on. Thus, the information they produce can be inaccurate, outdated, biased, irrelevant and even completely fabricated. Furthermore, AI can be used to purposely mislead or manipulate.
One way to mitigate these concerns is to provide assurance that there is a “human in the loop.” This is where generative AI is used to create content, but the results are reviewed, edited, independently verified and approved by one or more qualified humans before being published.
When asked if adding a label to indicate when generative AI was used responsibly with a “human in the loop,” 50% of respondents said it would make them more confident in the accuracy and reliability of the results, while just 16% said it would make them less confident. Interestingly, 34% said labeling content wouldn’t make a difference in their level of confidence.
Among respondents who are already confident in the accuracy and reliability of AI generated content, 73% said having a “human in the loop” label would make them even more confident in the content produced by generative AI, compared with only 3% saying it would make them less confident. On the flipside, among respondents who are not confident in AI-generated content, only 25% said a “human in the loop“ would make them more confident. Surprisingly, 30% of this group said including a “human in the loop” label would actually make them even less confident in the accuracy and reliability of AI-generated content. Furthermore, almost half (45%) of this group said having the label wouldn’t change their level of confidence.
These results highlight a major part of the AI divide. Those who are already favorable toward AI are the most receptive to strategies to improve the trustworthiness of content produced with generative AI, whereas those with a negative perception of AI seem to double-down on their dislike and will require a much greater level of convincing.
Can generative AI modernize legacy code bases?
Want insights on AI trends delivered to your inbox? Join the 451 Alliance.
