Meet Pi, the AI chatbot that’s bringing your personal assistant dreams to life! Designed by Inflection AI, founded by technocrats and entrepreneurs Mustafa Suleyman, Karén Simonyan and Reid Hoffman, the chatbot works by having users of the platform input their conversation and in turn receive advice or answers from a friendly, human-like AI.
The goal of Pi is to act like a “neutral listener,” probe users for more information, and then offer advice or answers related to the conversation – but only to the degree that a human could. Furthermore, Pi emphasizes its difference from a real person; when asked how it was doing, the chatbot responded that unlike humans, it wasn’t affected by stress, but rather was “programmed to be empathetic and understanding.”
By building on open-source large language models (LLM), the bot is trained to generate text responses by analyzing large amounts of data available online. And while Pi may not be able to truly replace your personal assistant, the chatbot can discourse on a wide range of topics, from offering recommendations for friends visiting New York City, to discussing your thoughts on graduate school —though it’s built to be autonomous, not to be constantly bombarded with questions.
Inflection AI highlighted its commitment to data privacy as well; noting that details that could reveal your identity are not used in the training. While users of Pi can delete accounts through the mobile apps or email —“PAUSE” response can be sent to the chatbot when wanting to temporarily stop hearing from it—Inflection AI is also actively working to improve its service by minimizing hallucination and redefining AIs learn and are trained.
It’s exciting to see companies like Inflection AI, who value privacy and automated intelligence, paving the way for the advancement of AI chatbots. pi provides us with yet another example of the potential that AI automation has to take mundane tasks like research and advice-seeking off our to-do lists.