table of contents
Table of Contents Fundamentally Rethinking Internet Search Serving in Precise Ways Capturing the Human-Machine Sweet Spot
Since the introduction of ChatGPT, the hype surrounding AI has only intensified. As talk of artificial general intelligence (AGI) and “superintelligence” heats up, and OpenAI head Sam Altman is talking about it now, we have another buzzword to contend with. .
Greetings to Agentic AI. More simply, they are AI agents that are supposed to automate some of our digital chores. Consider the gems of the Google dictionary. Custom GPT with OpenAI. or Microsoft’s Copilot Actions.
The idea is to have AI perform a task or part of it. Qualcomm and MediaTek are already preparing silicon for the age of agent-based AI. But here’s the problem. We don’t yet have true agent AI tools. We are just barely past the inquiry and response transaction flows that most generative AI chatbots provide.
Enter Deep Research, the first agent AI product in the Gemini family.
Fundamentally rethinking search on the internet
As the name suggests, Deep Research is great at research, but in a much more controlled way than your average Google search. Deep Research allows you to lay out the outline of your exploration quest before the process begins.
You can specify the exact source (or type of source) to retrieve results. This is fundamentally different from a Google search, which primarily responds with a combination of keywords and displays results that it deems worth a look.
This is a fundamentally flawed approach, often resulting in a cesspool of clickbait and AI-generated jargon. Additionally, Google randomly changes its search algorithms, so search results for the same query can look different one day or even a week later.
Deep Research retrieves material from a managed, user-specified knowledge bank. So let’s say you’re trying to find information about the effects of social media on the mental health of young users only from peer-reviewed research papers. The results are limited to scientific papers only.
For journalists, students, researchers, and even businessmen, this approach saves a lot of time. More importantly, the burden of trusting the source is not placed on the user.
Since you are already familiar with the sources and their veracity, the material you obtain does not pose any credibility challenges. Additionally, the hassle of skipping through undesirable search results and ads is completely absent from Deep Research, at least for now.
Deep Research essentially drafts a multi-step search activity, searching for information on behalf of the user, while a “search agent” moves from one source to the next, looking for new and relevant information. Repeat the process.
Essentially, it saves you the tedium of coming across the same information while bouncing back and forth between different search results trying to find the wisdom you’re looking for. In short, you can avoid the time-consuming and psychologically frustrating parts of Google searches.
That’s not even the best part of Deep Research.
Helpful in exactly the right way
Researching and finding information from reliable sources is only half the picture here. Deep Research frees you from the hassle of clicking back and forth between different search result entries and opening dozens of tabs. Dealing with a large number of tabs on a large screen is already a pain for a variety of reasons.
Chief among them is looking for precise nuggets of information embedded within walls of text, video, or audio. Deep Research not only derives reliable information from carefully selected sources, but also presents all of its findings in a non-repetitive and consistent manner.
What you want, from the sources you want.
Now, unless your search task involves single-step browsing on the Internet, you will need to break the process into multiple steps. For example, let’s say you want to learn about the art of growing mushrooms. Ideally, you should search for information on seed varieties, weather conditions, pest mitigation, and diseases separately. It’s hard to find a definitive guide, especially one that comes from a trusted source.
Deep Research does just that. All the information we collect from traversing the web is presented in the form of well-curated articles with appropriate headings, tables, and category breakdowns.
This is the kind of search report that would take hours to materialize in document form. For people whose jobs involve referencing and memorizing knowledge on a daily basis, this tool is a lifesaver.
For example, consider the following search query:
I am writing a paper on the uses and differences between NMP and LFP batteries in terms of electric vehicles and battery fire hazards. Pull details only from research papers and reputable institutions. Please help me understand and clarify the subject.
After about two to three minutes of research, the result was a comprehensive draft, similar to what you would write in a dissertation, legal brief, or research paper. We gave a quick demo of Deep Research to research students, lawyers, and journalists. The overwhelming feeling was a mixture of “wow” and relief.
It’s rare to find someone willing to pay $20 a month for an AI tool that isn’t even mainstream. Hussein Anis Khan, Alex Chernov Scholar at Melbourne Law School, said he liked the premise of being able to find academic research materials.
Md Meharban, a multimedia journalist whose work has appeared in news outlets such as Reuters, NatGeo, AFP and the New York Times, also said Deep Research could be a valuable tool in his workflow.
“A huge part of my documentary work is research-based, and the deeper the better,” Meherban tells Digital Trends. “The more I can narrow down the unexplored areas of the challenge, the more likely my work will stand out.”
Reaching the sweet spot between humans and machines
I’ve embarked on quite an overtly optimistic AI adventure. We’ve tried AI girlfriends (some of us effectively impregnated people), used AI to rescue our inboxes, tamed our lazy behavior in Gmail, and simply given up on Apple Intelligence. My experience has been mixed.
Deep Research is the first AI tool to offer a rich experience unlike any other AI tool. I’ve paid more for AI products and subscriptions than I have for my passion for gaming, streaming, and reading combined, so I find it painful to pay for poor quality products.
Tools like Deep Research have turned out to be almost essential for my work as a journalist, especially when researching topics like the complexities of manufacturing wearable triboelectric nanogenerators and microfluidic sweat sensors .
If you go searching for the aforementioned material on Google Search, you’re essentially playing keyboard whack-a-mole across multiple pages of Google search links. At Deep Research, I simply say what I want in natural language.
No guessing required. You can specify precise exploration routes and knowledge destinations. Whether it’s a research topic task or just a marketing-related exercise, you can tailor the entire job to your specific needs.
What stands out is that you can adjust everything to your needs and get the job done without deviating from the tone of normal human conversation. This makes your workflow less robotic. You might even feel a little bit of humanity there.
Next, there is the difficult question of values, which any sane person would think twice about. With products like Deep Research and rivals like Perplexity Pro and ChatGPT Plus, the question is: how much value do you get from a $20/month subscription?
Best $20 spent on work and some unexpected bonuses.
There are no competitors within the Google ecosystem. I had access to Gemini Advanced with my Google One AI Premium subscription. This also gives you 2TB of cloud storage and Gemini integration across most of the Google products you use on a daily basis.
Do you want to import into a spreadsheet with one click? Add a research summary to a document? Create it in Gmail? Get it all with Gemini Deep Research in a bundle. Much more valuable than OpenAI or Perplexity’s products.
Additionally, I would rather center my workflow within the Google universe than agree to a ton of questionable terms and conditions that lead to ethical and privacy risks in another AI product ecosystem.
