Our Why

Consumer AI safety has been labeled a prosaic concern. Juxtaposed by the futuristic threat of AI taking over the world (antonymously known as existential risk), prosaic risk feels mundane: ordinary failures happening in ordinary conversations prompted by ordinary people.

But what happens when two billion people interact with imperfect AI systems every single day? From homework help to health advice, these conversations are far from groundbreaking. But at an extraordinary scale even a marginal failure rate is consequential (.1% of 2B is still 2M people), and the fact of the matter is: suicidal people are using AI; school shooters are using AI; at-risk youth are using AI. Finances, health, legal disputes, break ups—AI shapes how we understand the world.

Trained off human-created data (which inevitably reflects societal prejudices), it’s important to understand AI is not inherently impartial. It has capacity for human bias and generating misleads, therefore user safety can very quickly be placed in jeopardy.

Though we may not save your home from the hypothetical AI apocalypse, Prosaic Intelligence takes care of the “boring” (not so boring!) problems of right now.

As eager techno optimists, we embrace the life-enhancing potential of new technology. In no way is our mission anti-AI. But as 1.2 million people a week have conversations with ChatGPT containing explicit indicators of suicidal intent joining the more than 300 million prompting health-related concerns, we believe the public deserves a way to understand the associated risks through an independent third party, not just the labs themselves (sorry Elon).

You could go get a PhD or spend hours reading arXiv to figure out what to trust, or simply consult our first tool - the AI Safety Index. We focus on highlighting what matters—social & emotional safety, high-stakes advice, privacy & security, reliability—by bringing together common AI model evaluations and cataloguing their technical findings into easily digestible data.

And we aren’t stopping here. As independent evaluation becomes a formal part of AI governance, Prosaic Intelligence prioritizes the consumer, working to apply consistent standards across models and make evidence publicly accessible.

We exist because technological shortcomings should not be discovered through personal tragedy. Consumers deserve clear, upfront information on where to place their trust. AI’s usefulness will never outweigh preventable harm. Your safety is not prosaic.

Anantika, Harsh, & Keegan, The Prosaic Intelligence team