AI Chatbot Pi Offers Emotional Support and Advice — Here’s How It Works

• A new chatbot called Pi, launched by Inflection AI, offers personal advice and support.
• LinkedIn co-founder Reid Hoffman and DeepMind co-founder Mustafa Suleyman co-founded Inflection AI.
• The Pi is designed to be friendly, but it makes it clear to users that they can’t feel emotions.

There’s a new AI chatbot on the scene, and we want it to be personal.

Short for “personal intelligence,” Pi was released on Tuesday by Inflection AI. The AI ​​startup has his three co-founders. Karen Simoyan, the company’s chief scientist. and LinkedIn co-founder Reid Hoffman.

Pi is conversational and designed to provide emotional support. In the opening dialog, Pi tells users that it’s “useful, friendly, and fun” and encourages them to ask for advice, answers, or just say whatever comes to mind.

Suleyman told Bloomberg Pi that it was intended to be a “neutral listener,” asking follow-up questions and offering advice.

When I said to my Pi, “Life has overwhelmed me,” the chatbot prompted me to share more.

“A lot of people feel like they just want to be heard, and they just want the tools to reflect what they say to prove they were actually heard,” Suleiman told Bloomberg. said.

Similar to OpenAI’s ChatGPT and Google’s Bard chatbot, Pi is based on LLM (Large Scale Language Model). This means that the bot is trained to analyze large amounts of content available online to generate text answers.

In another conversation, I told Pi that I was thinking about going to graduate school and was worried about missing the application deadline.

Pi gave me advice on other options to consider and asked me how I could keep myself busy.

The bot is designed to provide human-like support and advice, but it is also revealed that it is not actually human.

When I asked Pi how he was doing, he replied: I am here to be a supportive friend even though I have a range of emotions, both positive and negative.

Suleyman told Forbes that his vision for starting Inflection AI is to create chatbots that can respond to real-world problems.

“When I embarked on this project a little over a year ago, I had a core question: What makes a great conversation?” he told Forbes.

One time, I asked Pi to recommend museums for a friend visiting New York City. Pai secured some autonomy by mentioning “my favourites” and included several museums “off the beaten path”.

In general, Pi kept our conversation flowing smoothly with follow-up questions. At some point I asked if we could resume the conversation. Pi said he could change the topic, but he couldn’t completely reset the conversation.

A spokesperson for Inflection AI later told me that users technically cannot “resume” a conversation with the Pi unless they use different login credentials.

Users can delete their account by using the iPhone or iPad app or by sending an email request to Inflection AI using the phone number used to register, the spokesperson said. Users can also request via email that Inflection AI delete their copy of the conversation.

According to Forbes, the Pi was trained on the latest data as of November 2022. Suleiman told Bloomberg that no personally identifiable details are used in the training.

The chatbot is currently available for free in web browsers or mobile downloads for iOS, prompting the user to add their phone number after 10 interactions, so the Pi will send a text message to the user can ask how you are doing. If the user wants to stop contacting the Pi, they can text the chatbot “PAUSE”. Pi is also available to DM him on Instagram and Facebook Messenger, and text him on WhatsApp and SMS.

Like any AI chatbot, the Pi can say things wrong or inaccurately. Inflection AI says it works to “minimize hallucinations.” Inflection AI also says it is “creating a new form of ‘boundary training’ that will redefine how AI learns and trains.” Users can flag problematic conversations in their web browser or app.