Video yükleniyor...
Video Yüklenemedi
Very few predictions in technology last 25 years. In 2000, only 2yrs into Google, Larry Page predicted the company's #1 priority a quarter century later. Watch this ~1min clip. "Artificial Intelligence would be the ultimate version of Google. If we had the ultimate search engine, it would understand everything... show more
235,916 görüntüleme • 9 ay önce •via X (Twitter)
33 Yorum
Even the statement “because almost everything is on the web” was pretty ballsy in 2000!
larry is a true visionary
kind of wild how he mapped the entire arc before the tools existed
thx for sharing, deedy! looks like he understood early that search was never the product, interpretation was. whoever maps human intent to information flow becomes the os of the economy. search was just the 1st interface primitive :)
Now that’s what vision, conviction, and execution mean.
This is why long term thinking wins. Everyone in 2000 saw Google as a search box. Page saw it as the first step toward machine intelligence.
The irony here is heavy. Page perfectly described the "answer engine" model we see today with tools like ChatGPT and Perplexity, yet Google hesitated to fully deploy it for years because it threatened their core ad business. It’s the ultimate Innovator’s Dilemma. He saw the destination clearly—an AI that "understands exactly what you wanted"—but the company got too comfortable selling the blue links instead of the direct answers. Also, hearing him brag about "6000 computers" is a hilarious time capsule compared to the massive clusters running H100s today. It really shows how much raw compute has scaled to match that original vision.
The "6000 computers" and "100 copies of the whole web" really puts it in perspective. Today's AI training runs consume millions of GPU hours. The resources caught up with the vision, but the execution gap remains wild.
Omg just discovered that Larry is the GOAT. We need more clips of him speaking
Watching this now feels surreal. As a data engineer, it’s clear how accurate that thinking was even back then. He wasn’t talking about hype. He was talking about data, compute, and the ability to understand intent at scale. That’s exactly what we’ve been building toward for years. AI didn’t appear suddenly. It grew out of better data systems, more compute, and smarter engineering. Sometimes the best predictions are just engineers clearly seeing where the data is going.
They don’t own the web and it’s not theirs to profit from. They’re an indexer, nothing more.
Based on what many people have said about how good Larry was in envisioning the future, pretty sure he would not have let Google stumble the beginning of the AI race.
That vision only fully materializes once intelligence is not just searchable, but attributable, composable, and economically grounded at the system level.
Hear me out. The A.I. was the plan all along by the elites. The internet first, then the search engine. These two were built to gather data to build the A.I. It’s all going accordingly if you think about it. Transhumanism and man as a multi planetary species. The fire rises.🔥
I think it's easy to forget how long Google's been working on this. It's a great reminder that true innovation takes time and dedication to a core mission.
Page called it in 2000 Everyone thought Google was search He knew it was always the AI foundation 6000 computers then Millions now The vision didn't change Infrastructure caught up
Watch elon 25 years ago....he nailed many
They have had the biggest llm forever and a day
Organize the world's information and make it easily accessible to everyone. What other mission statement can possibly get us to AI? Also what after AI? Solve for intelligence and use that solve everything else! Google and Deepmind's mission statements. All encompassing!
most 25-year predictions age like milk. this one aged like he had the roadmap and just needed moore's law to catch up
Have they been collecting our data with this in mind since the very beginning? Was that what the fine print told everyone who created a website available through Google?
It's interesting how Page's vision frames AI not as some distant sci-fi goal, but as the natural endpoint of search-leveraging the sheer scale of data and compute that Google was already amassing back then. The intellectual pull he describes is what keeps engineers hooked.. turning raw resources into systems that actually understand intent. We're closer now, but that gap between "incrementally closer" and "ultimate" still demands the same relentless curiosity.
Google was ahead this whole time. They got caught by surprise with ChatGPT’s release for text assistants.
Great visionary and probably the reason why Google was among few to survive the dot-com bubble.
The most accurate part isn't the AI prediction. It's "we can get incrementally closer." Most companies skip the incremental part and wonder why their AI initiatives collapse under real load.
AI has been talked about longer than google has been around and the first AI tech popped up around the time that google started
Where is Larry page now?
Omg, sounds like the ultimate search engine has a long way to go 😂
This vision only becomes real when understanding, memory, and intent are continuous systems, not queries, which is exactly where personal agents start to matter.
The AI browser will manipulate the result. They will hide more shit.
that 100% came true, and google can take a lot of credit for that. all the data LLMs are being trained on today was created over the last 20 years because the web was easy to access and popular, mostly thanks to google.
Larry nailed it, rare when tech predictions age well.
I miss him
!["One of the very confusing things about the models right now: how to reconcile the fact that they are doing so well on evals. And you look at the evals and you go, 'Those are pretty hard evals.' But the economic impact seems to be dramatically behind. There is [a possible] explanation. Back when people were doing pre-training, the question of what data to train on was answered, because that answer was everything. So you don't have to think if it's going to be this data or that data. When people do RL training, they say, 'Okay, we want to have this kind of RL training for this thing and that kind of RL training for that thing.' You say, 'Hey, I would love our model to do really well when we release it. I want the evals to look great. What would be RL training that could help on this task?' If you combine this with generalization of the models actually being inadequate, that has the potential to explain a lot of what we are seeing, this disconnect between eval performance and actual real-world performance"](https://image.24vids.com/tw-1993450075616690474/amplify_video_thumb/1993449024045932544/img/W-1TjOXcXeiUbHaH.jpg)
