AI Briefing: Falling trust in AI poses a new set of challenges

Marty Swant March 11, 2024

0 4 minutes read

AI Briefing: Falling trust in AI poses a new set of challenges

By Marty Swant • March 11, 2024 • 5 min read •

Ivy Liu

Despite all the hype around generative AI across the tech world, there’s a growing gap in positive consumer sentiment.

In the past five years, consumer trust in AI has fallen globally from 61% to 53%, according to the 2024 edition of Edelman’s Trust Barometer, while trust in AI in the U.S. declined from 50% to 35%. Respondents globally trusted tech overall (76%) considerably more than AI (50%) and were more likely to embrace AI when institutions manage it well compared to when AI is poorly managed.

The annual survey — conducted in November with 1,150 people in each of 28 countries — also found rejection of AI was three times higher in developed countries than in developing markets. Meanwhile, just 38% of Democrats trust AI, alongside 25% of Independents and 24% of Republicans. On the other hand, 45% of Democrats reject AI while 25% accept it, and on the Republican side, 58% said they reject it and a mere 15% accept it.

Other research highlights some of the reasons many don’t trust AI. In a new report from UNESCO, researchers said AI models including GPT-3.5 and Llama 2 have an “alarming tendency” to generate content based on stereotypes about race, gender, sexuality and cultural biases. Meanwhile, the Center for Countering Digital Hate found several popular AI image platforms generated election disinformation in 41% of researchers’ test runs.

Election-related issues continue to be a big concern for advertisers. According to a new report from Forrester, 82% of U.S. consumer marketers said they’re worried about marketing their brands during the 2024 presidential campaign. And while AI-generated misinformation is among the concerns, it’s not the only one: Other “headwinds” include inflated ad prices, evolving regulations and consumer sentiment.

While various tech giants look for ways at improving trust and safety efforts around AI, regulators are also moving forward with their own proposals. Last week, the Federal Trade Commission announced plans for new rules related to AI robocalls in an effort to protect consumers and businesses from various scams. AI also got a brief mention last week during President Joe Biden’s State of the Union address when he said bipartisan legislation aims to “harness the promise of A.I. to protect us from peril,” which includes a proposal to ban on AI deepfakes.

Trust was also discussed last week during the FTC’s annual PrivacyCon event, which included numerous experts speaking on a range of topics. One speaker, Stanford University researcher Jesutofunmi Omiye, noted that humans are inherently trusting of information, which leads many people to believe the answers they get from platforms like Google.

“The thing we need to understand and remember is that human beings are trust machines,” said Omiye said during the event. “We think when a computer says something it’s very accurate, and that’s why a lot of people have been scammed. … And that’s just because we think computers are very right and we almost blindly follow computers’ instructions.”

Prompts and Products: Other AI News

Last week, two key ChatGPT competitors gave their AI chatbots noteworthy upgrades. One startup, Inflection AI, brought its Inflection 2-5 model to its Pi chatbot, added a way to conduct web searches and revealed Pi now has 1 million daily active users. Another startup, Anthropic, announced the debut of Claude 3.
HP announced a number of new AI products across its hardware and software, including a new portfolio of AI-powered PCs and a new partnership with Nvidia.
Less than a week after Elon Musk filed a lawsuit against OpenAI, the startup used old emails as evidence to refute Musk’s recent claims. The published emails in OpenAI’s blog post suggest Musk understood the startup would need to be more than a nonprofit and that he was aware developing AI would require keeping OpenAI’s models private instead of open-source.
Adobe said it’s adding new generative AI features for its Adobe Express mobile app across iOS and Android.
The web design company Wix debuted new generative AI features for a chatbot that helps design websites in seconds. The updates arrived not long after GoDaddy added similar AI features for small businesses with the debut of its Airo tools.
Other marketing tech companies that added new AI features last week include Gupshup, which launched a new “Conversation Cloud,” and Amperity, which added new generative AI tools for its CDP platform.
Cooler Screens rebranded as CoolerX as it looks to focus more on its AI capabilities.
This week, the European Union is expected to vote on sweeping AI legislation known as the AI Act, which could have major ripple effects on AI efforts across Europe and worldwide.

Research and other reports: Beyond the 1s and 0s

In a survey of 200 CMOs conducted by Statista and Plus Company, 81% said they use AI for media planning but only 36% use AI for attribution. When asked about AI’s impact in the next two years, 13% expect a reduction in SEO roles, 21% predict needing fewer copywriters, and 23% think AI will lead to a reduced number of analysts. Meanwhile, 53% think it’ll lead to an increase in creative roles, and 41% think it will create new roles for prompt engineers.

Stories from across Digiday

https://digiday.com/?p=537441