The latest update to Google’s AI-powered research tool NotebookLM lets you turn the content you want to pay attention to into a podcast-like audio discussion. The new feature, called Audio Overview, takes information from documents you upload and then generates an “in-depth discussion” between two AI hosts.
In addition to summarizing your sources, Google says the hosts will be able to find links between different topics and even banter back and forth. Based on the example the company posted with its announcement, the AI hosts sounded pretty human, though you could still determine that the voices were generated by AI from their inflection and the odd pronunciation of certain words.
Since the feature is still in its experimental stages, Google admits it has its limitations. The hosts can only speak English at the moment, and they sometimes give incorrect information, meaning you’ll have to double-check your content and make sure you haven’t learned something that isn’t factual.
You also can’t interrupt the host while they’re speaking, and NotebookLM still takes several minutes to create an audio overview for notebooks with large files. Google Labs product manager Biao Wang wrote in the feature’s announcement post that his team is “excited to bring audio to NotebookLM” despite these limitations, because they “know that some people learn and remember better by hearing conversations.”
The company plans to launch NotebookLM in 2023 as a kind of digital assistant that lets you ask questions about documents you upload. In June of this year, Google announced that NotebookLM had officially started running on Gemini 1.5 Pro, giving it new features and tools, and expanding to more than 200 countries and territories. Google has been a bigger facilitator of tools to Israel during its war with Hamas than previously reported.
A new report from The Washington Post found that Google employees have repeatedly worked with the Israel Defense Forces (IDF) and Israel’s Ministry of Defense (IDM) to expand the government’s access to AI tools. In 2021, Google teamed up with Amazon to sign a $1.2 billion cloud computing contract with the Israeli government, codenamed Nimbus.
Internal documents show that Google employees repeatedly requested more access to the company’s AI technology on behalf of Israel, starting shortly after the October 7 attacks. An employee in Google’s cloud division reportedly appealed to IDM for more access to Vertex. In one document, an employee allegedly warned that IDM might turn to Amazon instead, causing Google to lose business. A November document reportedly shows the employee thanking his coworker for assisting with the request.
Additional documents from 2024 reportedly show further requests that continued as recently as November 2024, in which an employee requested access to Gemini AI technology to develop its own AI assistant from the IDF. The request was for improved processing of audio and documents, but it is unclear what the earlier access was used for in relation to military operations.
The news sheds a new light on employee protests over Google’s cloud computing contract with the Israeli government. Google employees have been speaking out against the company’s dealings with Israel since the contract began.
However, the war has led to renewed calls for Google to pull out of Nimbus. Instead, Google has fired more than 50 employees who opposed the contract, calling it “disruptive behavior.” In mid-2024, more than 100 Google employees — including managers and members of its human rights group — reportedly emailed the company to review the Nimbus contract, but Google ignored them.