Google LLC is reportedly developing a slew of generative artificial intelligence features that can perform tasks such as giving life advice and creating budgets.

The New York Times reported the development effort today, citing internal company documents. The initiative is reportedly led by Google DeepMind, the search giant’s AI research group. The group was formed in April after the company merged two earlier AI teams known as DeepMind and Google Brain.

The search giant’s engineers are reportedly working on at least 21 different generative AI capabilities. One of the main highlights is a feature that can reportedly provide advice about “challenges in people’s lives.” A sample prompt obtained by the Times suggests the feature can process complex user queries containing several lines of text.

In December, Google’s AI safety experts are said to have cautioned that machine-generated life advice could have a negative effect on users. According to the Times, it’s not yet certain that Google will launch the feature and the other generative AI capabilities detailed in today’s report.

Those other AI capabilities are expected to focus on a variety of use cases. One under-development feature is described as an AI-powered tutor, while another is designed to help users create planning documents such as personal budgets. Google DeepMind is also said to be working on an “idea creation feature” as part of the initiative.

The search giant has reportedly partnered with Scale AI Inc. a well-funded AI startup, on the development effort. Scale AI sells tools that companies can use to create training datasets and integrate large language models into their software. The startup received a $7.3 billion valuation after a funding round in early 2021.

According to today’s report, Scale AI has assembled multiple teams to help Google test its new AI features. Those teams reportedly include, among other staffers, more than 100 experts with doctorates. 

“We have long worked with a variety of partners to evaluate our research and products across Google, which is a critical step in building safe and helpful technology,” Google DeepMind said in a statement today. “At any time there are many such evaluations ongoing. Isolated samples of evaluation data are not representative of our product road map.”

Google is also pursuing a number of other generative AI initiatives in parallel. 

Earlier this week, The Information reported that the company is building a portfolio of large language models called Gemini. The models, which will reportedly be capable of generating both text and images, are expected to become available this coming fall. It’s believed they will roll out to users either in the form of a new chatbot or as an upgrade to Google’s existing Bard AI assistant. 

Photo: Google