Hey Reader,

Welcome back!

This week, we'll show you how you can use the best AI Deep Research tool ever for free, share with you a prompt that eliminates AI hallucinations, and break down a story that has the entire music industry in a panic.

All that and more below! Let's get right into it...

🪄 App of the Week: NotebookLM Deep Research

Use Case: Best AI Deep Research for Free

Link: NotebookLM (Free, Not Sponsored!)​

Why you should care:
​NotebookLM allows anybody with a Google account to build "Notebooks" from various sources and do all kinds of AI analysis on the gathered data.

In the past couple of years, it became arguably the best AI tool for research, and this week it just got even better with the new Deep Research integration.

Bottom line: this is the best Deep Research we've seen in any AI platform. ChatGPT, Gemini, Perplexity, Claude...this is the best implementation of this feature yet.

Here's how you can use NotebookLM Deep Research for free today:

1. Go to NotebookLM using the link above.
2. Sign in with a Google account.
3. Hit the "Create New" button to create a new Notebook.
4. Click the "Discover sources" button in the top right of the menu that pops up.
5. Select "Deep Research" in the dropdown box in the Sources menu on the left side of the screen
6. Type a prompt telling NotebookLM what you want it to search for.
7. Send the prompt!

After a few minutes (depending on the complexity of your request) NotebookLM will return with a list of potential sources for your Notebook.

Here's where the magic comes in: with other Deep Research features like ChatGPT's, you don't get any input in this step. ChatGPT will simply go out, find 10-100 sources, and then start running analysis on those sources. You never get to look at the sources and say, for example, "please don't use these three sources as they are outdated."

With NotebookLM's Deep Research, this step makes all the difference, allowing you to manually review the sources if you want before NotebookLM begins integrating them into your Notebook.

So whether you're a NotebookLM power user or this is your first time hearing about it, go check out this new feature today!

📰 News That Matters: You Can't Tell Whether Music is AI-Generated Anymore

Link: People can't tell AI-generated music from real thing anymore, survey shows | CBS News​

Why it matters:
​AI-generated music has been steadily improving in quality for years, but this week, we got data-driven confirmation that it has passed a threshold many thought was still years away.

It's official: most people can't tell the difference between traditional music recorded in a studio by human beings and songs generated by powerful AI systems.

CBS News published an article this week discussing the results of a recent poll conducted by Ipsos in coordination with streaming platform Deezer. In the poll, people listened to two song clips created by AI and one by humans, and the results were truly shocking.

More than half of the respondents said they felt uncomfortable not being able to tell the difference. Over half believe AI will flood streaming platforms with low-quality music, and nearly two thirds worry it will lead to a loss of human creativity.

But even with all that concern, listeners are still pressing play.

Deezer says AI-generated tracks are exploding in popularity on its platform. Back in January, just 1 in 10 streams were from fully AI-created songs. Ten months later? It’s more than one in three, roughly 40,000 AI-generated tracks streamed per day.

Whether the industry likes it or not, listeners are voting with their ears.

The era of AI music isn’t coming someday. It's already here.

📝 Prompt of the Week: Avoid Hallucinations and Get Accurate Answers

Prompt Formula:
​I need you to answer my following questions using ONLY the information from this page: [INSERT URL]

Quick Description:
​This is one of the best ways to avoid hallucinations when using an LLM like ChatGPT. By narrowing its focus to only the information you provide, you can ensure that your answers are cited and ChatGPT didn't just "make it up".

For example, let's say you have a question about OpenAI's privacy policy. If you just ask ChatGPT to explain it to you, it may pull information from unofficial sources on the web or its own training knowledge of outdated policies.

But if you use this prompt and include a link to OpenAI's constantly updated policy page (https://privacy.openai.com/), you will be sure to get accurate, up-to-date answers.

How to Use:

1. Open ChatGPT.
2. Copy the above prompt.
3. Paste it into the chat box.
4. Replace [INSERT URL] with the link you want ChatGPT to reference.
5. Send the prompt!
6. Begin asking questions after ChatGPT confirms it can view the URL.

If ChatGPT is unable to view the URL you sent for any reason, you can simply copy/paste the entire page manually and you will get the same results.

📽️ Video of the Week: A new era of intelligence with Gemini 3

​

Gemini 3 is here, and the benchmarks and early reviews are incredibly promising!

Google says this new Gemini model is the best AI model ever at multimodality, a.k.a. understanding text, images, audio, video and more.

As for how it compares to the competition: Gemini 3 crushed even GPT-5 Pro on Humanity's Last Exam, which is arguably the most challenging AI benchmark.

All signs point to this being the best AI model ever released.

So check out this 2-minute video that breaks down what is new with Gemini 3 and how you can access it today!

That's all for this week!

See you next Thursday.

Much Love,
Igor & The AI Advantage Team

​

Was this email forwarded to you? Sign up for the weekly AI Advantage newsletter to stay up to date on AI tools and prompts that actually matter. ​ Sign-Up

​

Follow The AI Advantage:

​​​​

Written By:​
​Igor Pogany - igor@myaiadvantage.com​
​Daniel Pierce - danielpierce305@gmail.com​
​© 2024 All Rights Reserved.