Subscribe to discover Om’s fresh perspectives on the present and future.
Has Search Become Just a Feature?
You can now use Claude to search the internet to provide more up-to-date and relevant responses. With web search, Claude has access to the latest events and information, boosting its accuracy on tasks that benefit from the most recent data.
Web search is available now in feature preview for all paid Claude users in the United States. Support for users on our free plan and more countries is coming soon. To get started, toggle on web search in your profile settings and start a conversation with Claude 3.7 Sonnet. When applicable, Claude will search the web to inform its response.
On a typically gloomy San Francisco morning, while trying to sync my Philips Hue lights with Apple HomeKit, everything about the future of search — and, by extension, the traditional internet—snapped into focus.
Instead of opening a browser and clicking through pages of results, I found myself simply asking ChatGPT about my HomeKit problems. The response wasn’t a list of links but a synthesis of solutions, presented with the clarity of a knowledgeable friend. When I needed clarification, I didn’t open a new tab — I just asked. The experience felt less like using a search engine and more like conversing with someone who had already read every manual and forum post about my specific problem.
Claude’s announcement will likely reinforce behavior that’s already becoming commonplace among OpenAI users. By integrating web search capabilities, both Claude and OpenAI are transforming search from a destination into a feature — potentially sparking the next great upheaval of the internet. When using OpenAI with web search, users quickly realize how limited and inefficient Google’s traditional search offering has become.
This is just like how I felt when I experienced Google for the first time—even before it had made it to the market. After that first meeting with Google’s co-founders, established search engines like Yahoo, Lycos, and AltaVista suddenly felt antiquated.
Search has evolved far beyond its 25-year legacy of Google queries that lead users through search engine-optimized content and outdated forum posts. In today’s artificial intelligence-driven landscape, where tools like ChatGPT have transformed user behavior, search has become more conversational and integrated into everyday tasks.
This is the question I posed back in 2023:
If ChatGPT and its brethren represent a significant shift in how we interact with digital information, then it means clear and present danger for Google’s revenue stream. Can the company afford to put that revenue stream at risk? Can it afford to be timid? Google has to quickly decide whether to be “Sniff and Scurry” or happy being “Hem and Haw.”
I’ve witnessed several transformative moments in internet history: when Mosaic first made the web browsable, when Google made it searchable, and when Facebook made it social. Each wave fundamentally reshaped how we interact with information. This current shift feels similarly momentous, yet distinct in its own way. I’ve explored this transformation extensively, both in previous articles and in my newsletter, CrazyStupidTech.
This shift matters more than you might think. Even the browser, that faithful window into the internet for the past three decades, is starting to feel like a relic. We’re moving from a document-centric web to something more fluid, where information flows naturally through conversation rather than being bound by pages or URLs.
For traditional media companies, this transformation must (and should) feel like watching a slow-motion earthquake. The Atlantic and its peers are signing deals with artificial intelligence companies, hoping to shape this future. But they’re repeating past mistakes, reminiscent of when they handed control of their content to Facebook and Google. These media organizations appear to misunderstand a crucial point: AI platforms don’t simply direct users to content—they absorb and transform it.
The atomization of information is unfolding rapidly. Artificial intelligence doesn’t just search; it synthesizes, contextualizes, and presents information in a user’s preferred format. AI agents fetch needed information, distill it and deliver it without requiring users to visit individual webpages. The traditional web — with its banners, pop-ups, and paywalls — increasingly feels like a relic from a less sophisticated era.
But perhaps most telling is how natural this all feels. The chat just works better than the old way of searching. It’s more human, more intuitive, and more useful. We’re not just witnessing a new feature being added to our digital toolset – we’re watching the emergence of a new way of interacting with information itself.
The implications are as jarring as the earthquakes that occasionally rattle San Francisco. Consider Chegg, the homework help platform now suing Google because AI-generated answers are undermining its business model. But Chegg is missing the bigger picture: Students seeking homework help aren’t even going to Google anymore — they’re turning directly to OpenAI’s ChatGPT and Anthropic’s Claude for answers.
With this evolution, we might finally be approaching what the internet was supposed to be all along: a tool for thought, not just a collection of pages. At least, that is how the optimist in me sees the future.
March 20, 2025. San Francisco
FROM ARCHIVES
12 thoughts on this post
Comments are closed.
have you tried perplexity.ai?
it even shows the sources..
I stopped using it once OpenAI made “web search” available inline. I really don’t find them better and spending additional dollars on Perplexity made no sense.
Two comments.
First, we now know why Sergey Brin is demanding everyone at his shop work 60 hour weeks (and isn’t even promising ice cream and fussball tables). They’re going down, and I won’t buy the stock.
Second, your point about the Atlantic reflects something I’ve said for decades. Publications are communities. They’re self-contained. They’re environments where people with common interests gather.
Google’s threat is clear and present, and if they had ability to try and figure out their path, they would be really rebooting their business quickly. Sadly, they don’t know how to “monetize” Ai search yet. I think they will eventually move to that reality.
Even in the UK I see this shift happening. Partly fuelled by the barriers that Google increasingly puts in the way of getting simple answers.
For example, trying to get the latest information on something is almost impossible. The “News” tab often shows stuff from days, months or years ago. You have to click on Tools > Any Time > [Select a time range]. There is no “Latest”.
Also, any search result now starts with a selection of videos, of varying length, often requiring Facebook access. Then usually a selection of Reddit answers, and so you scroll on to see if anything looks like it will actually answer your question. It turns out that clicking on the “Web” tab cuts out a lot of the cruft, but until it was suggested I had no idea what it did and no incentive to try it.
Finally, any results you do click on with likely be so designed around SEO that you have to scroll for ages to see if you can spot it – there’s no way to go to the section shown in the Google result precis.
So I for one welcome our new AI Search overlords…
Rupert
You are absolutely spot on. The sheer lack of quality in Google results and search interface depreciation is why they are going to fail drastically. They are asleep at the wheel.
It’s not even that pages are designed around SEO, it’s that nowadays most of the pages Google brings up are clearly AI generated, or at least written with a good chunk of AI content. eg Cooking recipes no longer just give you the recipe, but have a back story 5 minutes long. The actual ingredients list is hard to spot and the instructions even more so. I am going back to physical cookbooks. Much more user friendly. I didn’t used to be like this a few years ago.
I feel like my next step, as someone who literally started using Perplexity for work this week for the first time, that cueing AI engines is the next big expertise for me. Do others feel that it’s not so much that AI will replace tasks or worker or workers but it’s how well one can cue the AI to produce the desired fruit?
I agree with all that you say, but I think one thing that’s worth keeping in mind is that Google has one advantage over the likes of ChatGPT because of its ecosystem.
The ability to use Gemini and then tell it to save things to Keep as a note is very useful. I’m not a Google fan but if ChatGPT was to develop some integrations with apps it would magnify its useful even further. Apple is miles behind, but if it can get the new Siri to work with the apps I already use, they are on to something very promising.
Finally, I have a much better understanding of what AI (esp., ChatGPT and Claude) WILL be doing for me. Thank you. In my ‘bones’, I felt this was what was developing as ‘AI’. And you have confirmed it for me. Thank you Om! The fog has been lifted for me………
Thank you Dave. Hope you are well.
I’m struggling to understand how AI search results thrive long-term. I think a lot of the information on the internet is there because of the promise of revenue generation from publishing content – thanks to the search engine model. It’s gone downhill recently, sure. However, the information from AI responses depends on this content. If this new paradigm means less revenue from publishing free content, this would result in reduced incentive to publish, and hence less content for AI models to scrape and learn from. Does this mean the responses from AI become stale, or dated, over time? Or does web content become dominated by large wallets/politically driven agendas, ultimately swaying the AI search results to some narrative – not unlike what’s happening with some social networks today? Perhaps the model to encourage unbiased content publishing (for AI’s to learn from) hasn’t yet been determined – or I’m just missing it somehow.