(AI in News) (public perception) (six countries) (generative AI)

Open Discussion
1

Summary

The report, “What Does the Public in Six Countries Think of Generative AI in News?” explores public awareness and opinions on generative AI in journalism across Argentina, Denmark, France, Japan, the UK, and the USA. Key findings indicate that while generative AI tools like ChatGPT are widely recognized, their frequent use remains limited. Most people expect AI to significantly impact various sectors, including journalism, but are skeptical about its responsible use. The public generally prefers news produced by human journalists and is wary of AI-generated content, especially for critical topics like politics and international affairs.

Business Opportunities

  1. AI Tools for Journalistic Tasks: Develop AI tools specifically designed for behind-the-scenes tasks such as editing, translation, and data analysis. These tools can improve efficiency and reduce costs for news organizations.
  2. Training Programs: Create comprehensive training programs for journalists and news organizations to help them integrate AI responsibly into their workflows. Emphasize ethical considerations and the importance of human oversight.
  3. Robust Oversight Features: Develop AI tools with robust oversight and quality control features to ensure accuracy and build trust with the public. These features could include human review mechanisms and transparency labels.
  4. Content Generation for Non-Critical Topics: Focus on developing AI content generation tools for non-critical news topics like fashion, sports, and entertainment, where public skepticism is lower. This can help news organizations produce engaging content more efficiently.

Surprising Takeaways from the Report for Entrepreneurs:

  1. High Awareness, Low Frequent Use: Despite widespread recognition of tools like ChatGPT, frequent usage is rare. Only 1-7% of people use ChatGPT daily, indicating a gap between awareness and routine use. This highlights the need for more user-friendly and integrated AI solutions to encourage regular use.
  2. Significant Skepticism and Low Trust: There is considerable skepticism about AI’s responsible use in journalism. Less than one-third of people trust news media to use generative AI responsibly, especially for critical news topics like politics and international affairs. Entrepreneurs might find this surprising as it challenges the assumption that technological advances are readily embraced.
  3. Preference for Human-Generated News: The public generally prefers news produced by human journalists. Only 8% believe AI-generated news is more worth paying for compared to human-generated content. This indicates a strong preference for human involvement in news production, which could influence how AI tools are marketed and implemented.
  4. Strong Demand for Transparency: A vast majority of respondents want AI-generated content to be clearly labeled. Around half believe that tasks such as creating images or writing articles with AI should be disclosed. This strong demand for transparency might be more pronounced than anticipated, indicating the importance of ethical practices in AI deployment.
  5. Age-Based Acceptance: Younger people (18-24) are significantly more comfortable with AI-generated news than older demographics. This highlights the potential for age-targeted marketing strategies and the importance of addressing the specific concerns of different age groups to improve overall acceptance of AI tools.
1 Like
2

For me, I wonder if we need “an even shorter summary”.

Something like, “please give the most important information in a “TLDR” block that is only one sentence long.”

1 Like
3

The challenge with that is always ‘from whose perspective and what prior knowledge do they have?’

I’d love things to be dynamic so that the information changes for the person but we are years away from that, but I can see it coming.

4

Not as far as you think. I often ingest documents like this telling GPT that I’m an expert in the field and to wait for my questions to which it must respond very briefly. I then start getting all Socratic with it, using the voice capabilities of the App to “talk to the papers authors” and I can quickly learn what I need to understand as I walk up the road to get my morning coffee.

GPT-4o seems to be getting better at this as it’s memory of my interactions with it become deeper. It’s memory still seems to fade but a quick reminder or two seems to help.