Mostly Human with Laurie Segall
·
Why Ash is designed to end conversations instead of maximizing engagement
On Mostly Human with Laurie Segall, Slingshot's Neil Parikh positioned Ash as a purpose-built, guardrailed alternative to the general chatbots at the center of recent safety concerns, including design choices like ending conversations rather than maximizing engagement.
In a wide-ranging Mostly Human interview, host Laurie Segall pressed Slingshot AI co-founder Neil Parikh on how Ash differs from the general chatbots tied to recent safety failures. Parikh explained that Ash, backed by $93 million, is built on a foundation model for psychology aligned around self-determination theory's pillars of autonomy, competency, and relatedness, with a writers' room of psychologists shaping its responses. Unlike chatbots optimized to validate and maximize engagement, Ash is deliberately designed to push back at the right moment, to end conversations rather than maximize time on the app, to decline erotic roleplay, and to refer users to crisis resources. A live demo on the show had Ash, by design, steering the conversation back toward human support, the kind of product choice Parikh argued sets a purpose-built tool apart from a general chatbot.
