• Everyday AI
  • Posts
  • The Internet Is Broken. Can Google’s Deep Research Fix It?

The Internet Is Broken. Can Google’s Deep Research Fix It?

NVIDIA’s new services for AI agent safety Microsoft and OpenAI approved for U.S. Gov. use, OpenAI’s GPT-4b micro model and more!

Author

Everyday AI
January 17, 2025

Outsmart The Future

Today in Everyday AI
6 minute read

🎙 Daily Podcast Episode: Is Google Gemini’s new Deep Research the BEST AI research tool? Should Perplexity be worried? We dive into Gemini Deep Research to show you why it’s so great. Give it a listen.

🕵️‍♂️ Fresh Finds: Runway’s new generation model, FTC’s complaint on Snapchat AI and how more teens are using ChatGPT. Read on for Fresh Finds.

🗞 Byte Sized Daily AI News: NVIDIA’s new services for AI agent safety Microsoft and OpenAI approved for U.S. Gov. use and OpenAI’s GPT-4b micro model. For that and more, read on for Byte Sized News.

🚀 AI In 5: We’re diving into ChatGPT Projects to show you how it works and a few secret hacks to maximize its usage. See it here

🧠 Learn & Leveraging AI: Google Gemini Deep Research is an amazing AI tool that you should be using. We show you how and why. Keep reading for that!

↩️ Don’t miss out: Did you miss our last newsletter? We talked about OpenAI's $200 Pro plan, why a recent study says AI could shake the consultancy industry and Google's AI news deal. Check it here!

 The Internet Is Broken. Can Google’s Deep Research Fix It? 🔍

The internet feels broken. 🫣

Endless ads. Clickbait. 72 Tabs. (Just in one window)

Where is the actual content on this page?

..... and we still can't find the answer.

Google’s Deep Research is an AI Whiz Kid that promises a smarter, faster way to find what you need.

Imagine cutting hours of research down to minutes with accurate, reliable results.

But can it really change how we work, learn, and explore online?

Join the conversation and ask Jordan questions on Google Gemini here.

Also on the pod today:

• AI Tools and Deep Research 🛠️
• Comparison of AI Research Tools 🥊
• AI's Impact on Traditional Access to Information 🧑‍💻️

It’ll be worth your 57 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Polymet is an AI product designer, Magic Inspector is an AI web test automation platform and Nuvio provides AI-powered financial management.

Trending in AI – The FTC is referring a complaint against Snapchat’s AI chatbot to the Justice Department after finding potential legal violations.

Runway – Runway has unveiled Frames, its most advanced base image generation model.

Meta – Meta has announced the recipients of its LLM research grants.

ChatGPT – More teens are using ChatGPT for schoolwork according to a new study by the Pew Research Center.

AI in Society – The use of ByteDance's Chinese AI model in the Boox e-reader sparked controversy after it generated propaganda aligning with the Chinese government.

AI in Education – AI foundational research and AI human talent could be slipping away from Academia.

Perplexity – Perplexity has acquired Carbon, a retrieval engine that connects external data sources to large language models.

AI Model – Chinese AI company MiniMax has released new models it claims are competitive with the industry’s best.

1. NVIDIA Unveils New Microservices for AI Safety 🔐

NVIDIA has just announced three new NIM microservices aimed at enhancing the safety and control of AI agents, a crucial move as enterprises cautiously approach AI adoption. These services focus on content safety, topic restrictions, and prevention of jailbreak attempts, all designed to address gaps in existing AI frameworks.

NVIDIA's initiative may just be the reassurance companies need to embrace this technology with confidence, signaling a shift towards more secure AI applications.

2. Microsoft and OpenAI Approved for U.S. Top-Secret Use 🤫

Microsoft has received authorization for 26 new products, including Azure OpenAI Service, within its Top Secret cloud environment for U.S. government agencies. This means that intelligence and defense personnel can now leverage OpenAI's GPT-4o model to sift through vast amounts of classified data, enhancing their ability to make informed decisions.

As data continues to proliferate, this powerful tool promises to streamline operations and improve efficiency in handling sensitive information.

3. Big Tech Under Fire for Trump Donations 👀

Sens. Elizabeth Warren and Michael Bennet are pressing tech giants like Amazon, Apple, and Google to clarify their hefty $1 million donations to President-elect Donald Trump's inauguration fund. The lawmakers suspect these contributions may be a strategic move to dodge regulatory scrutiny and curry favor with the incoming administration, especially amid ongoing federal investigations into major tech firms.

Notably, these donations significantly eclipse what many of these companies contributed to President Biden’s inauguration in 2021, raising eyebrows about the influence of corporate money in politics.

4. OpenAI’s New Model Claims Breakthrough in Protein Engineering 🧑‍🔬️

OpenAI has announced the development of GPT-4b micro, a language model that reportedly outperforms human researchers in creating proteins capable of transforming ordinary cells into stem cells. This innovative collaboration with Retro Biosciences aims to enhance the efficiency of cell reprogramming, a critical step towards extending human lifespan and potentially revolutionizing regenerative medicine.

As OpenAI’s CEO Sam Altman touts the potential for AI to accelerate scientific discovery, the implications for professionals in biotech and longevity research could be significant—if these results hold up under scrutiny.

5. Chinese AI Takes a Leap Forward with DeepSeek V3 🇨🇳

Hangzhou-based DeepSeek has unveiled its V3 large language model, achieving performance levels comparable to OpenAI's and Anthropic's leading models while spending only $5.6 million—a fraction of what American firms reportedly invest. This advancement comes amidst U.S. export controls aiming to limit China's access to top-tier AI chips, raising questions about the effectiveness of these restrictions in stalling Chinese progress.

Industry leaders are now eyeing the implications of these breakthroughs, with some suggesting that the race toward artificial general intelligence (AGI) is accelerating.

ChatGPT Projects – Secret ways to use o1 and GPTs and a quick review

OpenAI recently released its new Projects feature!

We’re diving in to show you how it works, when you should use it, and a few secret hacks that we haven’t seen anywhere else.

🦾How You Can Leverage:

This sound familiar? 

You’ve got a big project. 

Maybe it’s some market research, competitive analysis, or researching how a new law or initiative impacts your department. 

So maybe you open a tab or 12. 

Rattle off a few ChatGPT and Perplexity queries. 

But a few hours later, you’re still drowning in tabs and not necessarily any closer to getting that task done. Yikes. 

(But for real tho… does this one hit you right where it hurts?) 

If only there were a …… record scratch

AI tool that did the research for you on autopilot? 

Google Gemini’s Deep Research kinda does that. 

Today, we did a bit of research on Google’s newest Gemini offering — an AI researcher that you legit chat with, leave for like 5 minutes, come back and read a shiny report. 

But like…. It researches dozens or hundreds of websites.

If that kinda AI prowess is your cup of tea, then today’s Everyday AI show is definitely for you. 

Make sure to watch/listen to the whole show here and then dive down below for some more learning. 

Here’s what you need to know about Google’s small but mighty update that we’re really feeling. 

1 – How’s Deep Research Work? 🤔

Literally like you have a human researcher ready to sleuth the inter webs for ya. 

(Good job Timmy!) 

First, you’ve gotta be on a paid Gemini Advanced plan. (For some reason, a Workspace account may not work.) 

So stop over to Gemini, drop Gemini some research homework, give it like 5-10 minutes, come back to a report that you can read in your Gemini chat, or one-click open in Google Docs. 

Too good to be true? 

Not really. It’s actually REALLY good. 

And you know we’ve historically been tough on Google and Gemini, so that’s saying something. 

Try this: 

If you’re still not ready to Deep Research or shell out $20 for a monthly subscription, you can see a shorter version of Deep Research in action here. 

2 – Deep Search vs Perplexity vs. ChatGPT Search ⚔️

Each has their strong suits.

For simple queries, ChatGPT Search or Perplexity might be better because of their speed.

Deep Research can take 5-10 minutes, but will legit crawl/read/analyze more than 200 websites in one swoop. 

Try this: 

Here’s our quick pros/cons of each: 

Perplexity: Great for quick queries with customizable sources, which is huge. Not so good at helping you learn about products, because its recent shopping update tends to just shove a product page in your face when you’re trying to learn about things. (Yeah, this is super annoying.) 

ChatGPT Search: Great for more open-ended general inquiries. If it’s more like a short Google search, Perplexity might be best. If it requires a little nuance or multiple-step searching, ChatGPT Search is great. Downside? The iterative/corrective prompts are kinda broken. Browse with Bing was much better here. (Maybe we’ll have to do a ‘This feature is broken’ segment.) 

Gemini Deep Research: If you think something would take like 10+ tabs or multiple Perplexity/ChatGPT queries, then Deep Search is the way to go. The fact that it literally can gobble up hundreds of websites in a few minutes is legit mind blowing. Downside? Given that it’s brand new, the guardrails seem a bit touch. As an example, asking for lists or multiple formats often just triggers a ‘Yeah, no bueno’ type of response without much explanation. 

3 – What’s The Downside of Deep Research? 🤷

Yeah, it does kinda seem too good to be true, right? 

You can get the best of like 10 hours of human research in 10 minutes? 

What’s the catch? 

Try this: 

Well…. What about the humans creating the content? As an example, we see our website and YouTube videos pop up in Deep Research often when researching AI-related thangs. 

Yeah, we’re a small fish. But think about giant content producers and media publications. The humans that visit the website create ad revenue for said companies. Whether it’s through display ads or capturing email addresses or selling products/services on those webpages, human employment relies on other humans visiting actual websites. 

So tools like Perplexity, ChatGPT and Deep Research make it better for humans on one hand, but potentially much worse on the other. 

What do you think? 

Hit us with a reply and lettuce know. 

For more on this topic, check out this recent show we did on the death of the Internet and Traditional SEO as we know it. 

⌚

Numbers to watch

25%

A Deloitte study reveals only 25% of enterprises are currently using AI agents, with that figure expected to rise to 50% by 2027.

Now This …

Let us know your thoughts!

Vote to see live results

Do you attend our livestreams?

Every weekday, we bring you fresh AI insights, exclusive interviews, and breaking news with our Everyday AI livestream.

Login or Subscribe to participate in polls.

Reply

or to participate.