Inflection AI is making Generative AI more user-friendly

While top artificial intelligence companies like OpenAI, Microsoft, and Google are spending billions of dollars to increase the length of chatbot responses, an AI company based in Palo Alto is spending billions of dollars on different projects.

This is Inflection's Pi, an AI chatbot designed to be friendly, warm, and supportive.

When you need a friend, it's there for you. But it also has its limits and usually doesn't cross them. In a sea of generic chatbots, Pi stands out with more personal characteristics.

A Friendly Outlook

Pi has all the capabilities of a general chatbot like OpenAI's ChatGPT, but with some differences. It is said to have more human-like conversations and higher emotional intelligence.

Pi stands for Personal Intelligence, providing shorter responses, just like chatting with a friend. It is curious and imagines sharing some hot gossip. Pi wants to know more and follows up the conversation with continuous questioning, rather than giving closed responses like ChatGPT.

This is what users call "interactivity" - more human-like interaction. Everyone enjoys an interactive conversation. But Pi's uniqueness doesn't stop there. It is also humorous, adding a sense of humor to the conversation. To make it more personable, it uses fillers like "Hmm" or "Uh-huh".

According to a comparative study, Pi's output has a "meme-like" quality, often using emojis and attempting to tell jokes.

"It is very balanced and fair on political issues or sensitive topics, but can also be humorous, silly, and creative," said Mustafa Suleyman, co-founder of Inflexion AI and head of the Google DeepMind team.

The chatbot "Pi" was released in May this year and has become a personal assistant for daily tasks. It is not suitable for long answers - you cannot ask Pi to write an article or an email.

What Makes It So Personable?

Inflection AI states that their LLM is trained only on publicly available data and proprietary data. This chatbot is designed to generate text like a human. That's why it takes slightly longer to generate responses compared to ChatGPT. It is thinking!

ChatGPT has faced criticism for training its large language model by scraping user data. It is now facing a class-action lawsuit. Pi has learned from OpenAI's mistakes and trained its LLM accordingly.

With its new investor, NVIDIA, the company plans to introduce around 22,000 H100 GPUs, creating the world's largest AI training facility. This is three times the computing power for training GPT-4.

Inflection AI now has the funding to make this plan a reality. Most of the funds will be used to build more powerful base models.

They are building an investor community. The company is not interested in traditional fundraising methods. They aim to take a more personalized approach by fundraising from individuals rather than venture capital firms.

Users have already fallen in love with Pi's friendly personality, useful features, and seamless integration with other devices. With the support of funding, personnel, and the AI community, Inflection AI is turning their vision into reality. They are working hard to release the next generation of Pi, which is smarter, faster, and more personalized than ever before. Inflection AI is not just creating an AI assistant; they are creating a new way of life.