AI’s Hard Takeoff: Deep Learning’s Impact
10:05 Share

Welcome back, everybody. It feels like every day there's a new breakthrough, some mind-blowing application of AI that makes you wonder, like, what's even real anymore?

And lately a lot of that buzz has been around this thing called deep research from OpenAI. It's causing quite a stir online to say the least. Now I know AI can seem like this big intimidating topic, but trust me, you don't need a computer science degree to understand the incredible things that are happening. In fact, I want to start with a personal story that really hit home for me. It shows how AI is starting to touch our lives in the most unexpected and even emotional ways. So this story is about Felipe Millan.

He's an employee at OpenAI, and he recently shared his experience using deep research to help make medical decisions for his wife. She had undergone a double mastectomy for breast cancer, and they were facing this really agonizing decision about whether or not to go through with chemotherapy. Imagine the weight of that decision. They were getting conflicting advice from different doctors, and the emotional roller coaster was just overwhelming.

It was in that moment of uncertainty that Milan turned to deep research. He input all of his wife's medical information into the system, and the AI went to work. It scoured hundreds of medical studies, even uncovering research that Milan himself, with all his expertise, hadn't come across. And then it generated a personalized report tailored specifically to his wife's situation, weighing the potential benefits and risks of chemotherapy in her particular case. What's really remarkable about this story isn't just the technology itself.

but the profound impact it had on a personal level. This AI tool provided Milan and his family with a level of clarity and confidence that at that moment, human doctors couldn't match. It empowered them to make a truly informed decision during an incredibly difficult time. You know, this experience really highlights how AI is no longer confined to theoretical discussions or research labs. It's stepping into our lives, even influencing life-altering medical decisions.

And this is just one example of how AI is changing the way we access and process information. But the story of deep research doesn't end there. It's also sparked a revolution in AI development itself. And that's where things get even more interesting.

See, while OpenAI might have been the first one to really make a big splash with these reasoning models, another company called DeepSeq decided to kind of shake things up. They made their models open source, meaning anyone could access the code and actually see how these powerful AI models worked. It was like throwing open the doors to this secret laboratory and inviting the whole world to come in and experiment. This move triggered something called the DeepSeq effect, where AI development basically went into hyperdrive.

researchers and developers from everywhere could now collaborate and build upon each other's work and push the boundaries of what was possible much faster than before. And here's a cool little anecdote that illustrates this perfectly. Simon Willison, he's a pretty well-known developer, challenged DeepSeek's AI to find ways to improve its own speed. And believe it or not, it actually did it.

The AI analyzed its own code, identified areas for optimization, and boom, made itself twice as fast. It's like we gave AI the tools to become its own teacher, constantly learning and evolving at this crazy rate.

And even Sam Altman, the CEO of OpenAI, who's known for being pretty bold with his predictions, admitted that there's now a straight shot to AGI, artificial general intelligence. AGI, for those who are familiar, is that point where AI can basically perform any intellectual task that a human can. It's the kind of thing that used to be pure science fiction, but now it feels like it's right around the corner. And this brings us to a really interesting observation by Ethan Mollick.

He's a professor who specializes in AI, and he argues that we've moved from the era of searching for information to the era of AI conducting research for us. Think about that for a second. Google was amazing for finding information that already existed, but deep research is actually generating new knowledge, pushing the boundaries of human understanding.

Mollick shared this anecdote about giving deep research a pretty complex prompt, and within minutes it produced an analysis that he described as PhD level, surpassing what even human experts could come up with using traditional methods. To really visualize how fast AI capabilities are growing, I want to show you a graph that tracks AI performance on STEM questions, science, technology, engineering, and math.

This graph shows what's called a hard takeoff in AI, where progress suddenly accelerates at an unprecedented pace. The blue line on this graph represents human PhDs using Google to answer those same questions. So these are experts in their fields armed with the most powerful search engine in the world. And guess what? AI has already surpassed them. This raises a pretty important question.

What happens when AI starts applying its research skills to improving itself? That's the essence of the hard takeoff, a point where AI development could explode exponentially, potentially leaving human capabilities in the dust. And the reality is, this isn't some far-off futuristic scenario.

Deep research is already revolutionizing fields way beyond just tech, from medicine and science to even law publishing education, you name it. It's like this Swiss army knife of knowledge that can be applied to practically any area where research and analysis are important.

Take medicine, for example. Daria Unumass. He's a biomedical scientist who specializes in aging and cancer immunotherapy at Jackson Lab. It's called deep research a total game changer, and he's not just hyping it up either. He talked about giving deep research two different cancer cases, one that was in his area of expertise and one that wasn't. He was blown away by the reports it generated, describing them as simply impeccable, like something a specialized MD would write.

In fact, he even said that deep research surpassed the insights of some specialized doctors in certain areas. But the impact of deep research goes way beyond just diagnosing individual cases. Sam Altman has claimed that it can already do a single-digit percentage of all economically valuable work globally. That's pretty mind-blowing when you think about it. And he believes that deep research is on the verge of not just synthesizing existing knowledge, but actually inventing brand new knowledge.

Now, in the midst of all this excitement, there's also a little bit of drama unfolding, and this is where things get kind of funny. Remember how I mentioned that DeepSeek's open source approach really shook things up? Well, it seems that some folks over at Google are feeling a bit inspired by OpenAI's deep research. Some Google employees have been complaining, accusing OpenAI of not only copying their functionality, but also stealing their name.

It turns out Google already has a research project called Deep Research, and they're saying that OpenAI is trying to capitalize on their brand recognition. One Google employee, Arush Selvan, even reposted OpenAI's Deep Research announcement with a comment, Well, I wonder where they got that idea from. And Logan Kilpatrick, the lead product manager for Google AI Studio, chimed in with, Product experience for Deep Research is pretty great. Try it out.

subtly pointing people toward Google's version. I'm not going to pick sides in this little spat, but you have to admit, the irony is pretty funny. Google the giant of search accusing someone else of being a copycat. It just goes to show how much deep research is disrupting things, forcing even the biggest players to change their game. But hold on to your hats because there's one last twist in this whole story that makes you wonder what's next in the wild world of AI. Sam Altman, in his usual cryptic way, hinted that deep research isn't even the big thing he's been teasing.

He said, deep research is not the one more thing and that there's a stronger one to come. I don't know about you, but that leaves me with a mix of excitement and honestly, a little bit of nervousness. What could possibly be bigger than deep research? What other wonders or maybe even anxieties are waiting just around the corner? Yeah, I guess we'll just have to wait and see what Allman's got cooking. But in the meantime, all of this has got me thinking about a bigger question. You know, like, what does it even mean to be human in a world where AI is becoming this powerful?

It's the kind of question philosophers and sci-fi writers have been debating forever. But it feels especially relevant right now. As AI starts to blur the lines between what we thought were uniquely human capabilities, we're kind of being forced to confront our own limitations and redefine what makes us special. For centuries, we've seen ourselves as the most intelligent species on Earth at the top of the food chain. But now we're creating machines that can outperform us in tasks we used to think were only ours.

like playing complex games, composing music, writing poetry, even doing scientific research. It's a bit humbling, honestly, but it's also a huge opportunity, an opportunity to take a step back and ask, what are the things that truly set us apart? Where do our strengths lie in a world where AI is mastering these skills we used to consider the hallmarks of intelligence? One thing that really stands out to me is our capacity for empathy, for compassion, for emotional intelligence. These are qualities that, at least for now, AI can't replicate.

And as AI gets more advanced, I think these uniquely human traits will become even more valuable. We'll need that emotional intelligence to navigate all the ethical dilemmas that AI is going to throw at us, to build meaningful relationships in a world with both humans and machines, and to create a future that benefits all of us, not just the people writing the code. It's interesting to think of AI as a sort of mirror, reflecting back to us both the best and maybe the not-so-great parts of humanity.

On one hand, AI has this incredible potential to amplify our creativity, our ingenuity, our drive to solve problems and make the world a better place. But we can't forget that AI can also amplify our biases, our prejudices, even our capacity for making really bad decisions. It all comes down to how we choose to develop and use this technology. And that brings us back to those ethical considerations we've been talking about. We need to be incredibly thoughtful about how we build and use AI, making sure there's transparency, fairness, and a focus on what's best for everyone.

The future of AI is inseparable from the future of humanity, and it's a future that we all have a responsibility to shape consciously and ethically. So we've covered a lot today from these really personal stories to these mind-blowing breakthroughs that almost feel like science fiction. And it's clear that AI isn't just some passing tech fad. It's this force that's changing every corner of our world.

So as we wrap up, I want to leave you with this. Don't just sit back and watch this AI revolution happen. Get involved. Learn about what's going on. Talk to your friends and family about it. The future isn't something that just happens to us. It's something we create together. And the decisions we make today will shape what that future looks like. So stay curious, stay informed, and most importantly, stay engaged.