TIME’s cover this week featured the OpenAI co-founders with the headline “Trust Us”. Diane Brady’s CEO Daily yesterday spotlighted Meta’s need to rebuild trust. In a major essay and interview with The New York Times Bill Gates questioned whether leading AI developers are sufficiently communicating the full extent of AI’s potential impacts on society. Trust, it seems, may be a new kind of singularity in the race to develop and deploy AI.
Our latest research sheds light on the public’s perspective on the topic. 72% of Americans support pausing or slowing AI development so each model release can be evaluated more carefully. Among people who use AI every day that number climbs to 78%. Crucially, 72% of daily users say a slowdown would increase their trust in AI models.
What would people want addressed during a slow down or pause? Data privacy and security leads at 21%; AI safety — the risk of losing control, or of AI causing harm — comes in at 20%, followed by the accuracy and reliability of outputs at 16%. Job loss and economic displacement (10%), and broader concerns about societal overreliance on AI (10%) also feature.
One way for companies to build trust is through greater transparency. We analyzed 1,002 of the largest US-based firms across 9 critical AI safety and governance areas selected based on their importance to the American public. Let’s just say we found much room for improvement. Nearly half (498) disclose on none of the key issues. Only 10% of companies have published anything on preventing harm, deception, or manipulation; 14% commit to human oversight. Board oversight of AI is the most common disclosure, at 26%. Salesforce is the only company of the 1,002 to disclose across all 9 criteria. Maybe its partnership with Anthropic announced this week will help build trust for the latter ahead of its expected IPO later this year.
New capabilities released this week within Just Intelligence now enable companies to see where they stand on the public’s AI priorities, how they compare to their peers, and examples of what good really looks like. We’d love your feedback.
Be well,
Martin
71% of adults think AI will lead to fewer jobs in the United States over the next two decades. Explore the rest of Pew Research’s findings here.
Reuters looks at how Meta’s plan to replace nearly two-thirds of its workers with AI backfired.
Bill Gates released an essay on his worries about AI, and the need to prepare business and society for rapid change.
The Wall Street Journal reports that Google has moved its “AI responsibility” team out of its DeepMind Labs, with members of the team worrying about the impact this will have on their work.
The New York Times reports that Meta will have to pay $17 billion in a landmark settlement around its role in social media addiction among children.
CBS reports that the functional unemployment rate – which includes people only able to secure part-time work when they are seeking full-time work – stands at 25%, and growing.
Investopedia looks at the viral debate over a $20 burrito to discuss U.S. inflation, employment trends, and the constant feeling Americans have that everything is too expensive.
This chart comes from our newest research, and shows that 72% of power AI-users support pausing or slowing AI development so each model release can be evaluated more carefully. Explore the full findings here.
