Google’s secret weapon for work
Google’s NotebookLM AI platform can be a fun tool to use in the office and can even transform plain pieces of text into a conversational podcast.
Large language models like ChatGPT have completely revolutionised how people perform their digital tasks, especially in the office.
From summarising long pieces of text to writing birthday messages for colleagues, there seems to be nothing artificial intelligence programs can’t do.
As a result, most of the major tech companies have built their own LLMs. These include Microsoft’s Copilot, Meta’s Llama, and Google’s Gemini.
However, Google’s experimental NotebookLM does things a little differently.
Instead of the typical way a large language model works, where a user interacts by asking it a question, and it responds by summarising, translating, or predicting text in a human-like way, NotbookLM allows users to ground the model in their own sources.
Google explains that NotebookLM allows users to let the model become well-versed in the information relevant to them. It means this literally.
To use NotebookLM, users need to upload a source. This can be any piece of text, YouTube video (which must have a transcript), audio recording, or website.
For instance, if someone is writing an essay about the pros and cons of LLMs, they can upload multiple sources to provide them with the most objective perspective possible.
However, comparing these sources can take ages, and reading and listening to all of them can be tiresome.
Instead, users can simply tell NotebookLM what information they need and present it from all the sources at once, providing references so the user can double-check it.
Users can also ask the model to point out where sources contradict each other or points of disagreement.
In essence, NotebookLM can provide users with insight into multiple sources simultaneously.
One of the most notable features is the ability to create an AI-generated podcast directed by the user.
For example, suppose a source argues that LLMs should undergo testing before they are released to the public, and another says this is unnecessary. In that case, NotebookLM can create a podcast with arguments from both sides.
While these features may seem like the next best thing, it is essential to note that the program is still in beta mode and undergoing testing.
When using it, users should always double-check what the LLM has produced.
Loading ...