Google’s Project Ellman: Combining pictures and search info to make a digital twin chatbot

  • Project Ellman Overview: Google is exploring “Project Ellman,” using Gemini AI models to create personalized life stories for users by analyzing data from Google Photos and Google Search.
  • Multimodal Large Language Model: The project is powered by Google’s Gemini, a new multimodal large language model designed to process diverse data types, including images and text.
  • Biographical AI: Project Ellman employs AI to generate biographies based on personal data, analyzing photos, internet information, and more to provide a comprehensive understanding of a user’s life.
  • Data-Driven Personalization: Users can interact with a personalized chatbot akin to ChatGPT, equipped with a deep understanding of their life, enabling more meaningful and tailored conversations.
  • Privacy Considerations: The article raises questions about user privacy and the extent of data access required. Google emphasizes responsible development, prioritizing user privacy as a top concern in the exploration of this concept.
  • Predictive Capabilities: Besides creating life stories, Project Ellman could predict user interests, potential purchases, travel destinations, and preferences by analyzing saved images, contributing to targeted advertising strategies.
Google's Project Ellman: Combining pictures and search info to make a digital twin chatbot
Google’s Project Ellman : Combining pictures and search info to make a digital twin chatbot (Image: Getty Images)

Google is reportedly considering leveraging its latest Gemini AI models to analyze and amalgamate information from Google Photos and Google Search to construct a personalized life story for users.

The endeavor, currently under the moniker “Project Ellman,” is set to utilize Google’s newly introduced multimodal large language model, Gemini. The goal is to aggregate diverse data types from various sources, such as photos stored on Google Photos or publicly available information from the internet, to develop a more individualized chatbot.

Project Ellman: Google’s Ambitious Venture into Personalized AI Biographies

Staff members from Google Photos and Gemini introduced Project Ellman, envisioning the potential product as a chatbot well-versed in users’ life details. According to CNBC, the project was described as follows: “Imagine opening ChatGPT, but it already knows everything about your life. What would you ask it?” The project is supposedly named after the literary critic and biographer Richard David Ellmann, known for his works on Irish writers like James Joyce, Oscar Wilde, and William Butler Yeats.

Under Project Ellman, AI would be employed to craft a user’s biography based on their personal data. Google explained, “We can’t answer tough questions or tell good stories without a bird’s-eye view of your life.” The approach involves scrutinizing photos for tags and locations to identify meaningful moments, with the overarching goal of understanding a user’s life comprehensively.

“We trawl through your photos, looking at their tags and locations to identify a meaningful moment,” a presentation slide reads. “When we step back and understand your life in its entirety, your overarching story becomes clear.”

For users with memory lapses, Ellman Chat could assist in answering queries about past pets by analyzing data, such as photos containing animals or family members next to pets. Google refrained from disclosing the level of access users would need to grant the model to collect personal data, stating that it is a concept in the early exploration stages. A Google spokesperson emphasized the commitment to responsible implementation, with user privacy as the top priority.

Gemini, the AI model at the core of Project Ellman, has the capability to recognize significant life events by analyzing photos, such as graduations or vacations. It could also piece together details about a person’s educational background or places visited by analyzing information available on Google Search. Google illustrated the process of enhancing a detailed view of someone’s life by adding personal data.

The presentation highlighted the power of the large language model (LLM) in taking unstructured context from various sources to understand different aspects of a user’s life. An example provided demonstrated how the LLM could infer details about a person’s birth and family relationships by leveraging knowledge from different parts of the data.

Beyond creating biographies, Project Ellman could potentially predict users’ interests in products or travel destinations by analyzing saved images. It could also identify users’ most frequently visited websites and apps, contributing valuable insights to the advertising landscape. Google assured that the project is in the early stages of exploration, emphasizing a commitment to responsible development and user privacy protection.

Read Original Article on The Register