There was such a development in the field of artificial intelligence last week that I cannot see anything more important in the world right now. This development is Open AI's Deep Research, I am not talking about DeepSeek developed by the Chinese here, but this Deep Research has a lot to do with that DeepSeek. Deep Research has obviously used certain information in the article explaining DeepSeek. I don't know who is copying from whom anymore, but Deep Research is groundbreaking in every way. When you read what I have explained in this article, your perspective on the future, your perspective on your career, your perspective on the future of your children, your perspective on the future of the world will completely change. You will say, where is humanity going? Because humanity seems to be taking a final exam and the possibility of us losing to artificial intelligence in that exam has increased a lot. What does Deep Research developed by Open AI mean? Why is it so controversial? Why is it so revolutionary? How did cancer diagnosis treatment start to be used in practice right now? Although it has only been on the market for a week, I will tell you what experts say about it.

As you already know, Artificial Intelligence is not my area of ​​expertise, I focus more on stocks and markets, but developments in artificial intelligence will change the world so much that I also have a lot of investments in the field of artificial intelligence. Let's start with what Deep Research is. It is a new artificial intelligence model developed by Open AI. It actually includes advanced logic models. In other words, it can do reasoning. It also has agent-based systems called Agentic Systems. This new system has complex thinking skills and can also take action. For example, it can scan the internet, write code, run codes, and integrate with other tools. As a result, it does research for you. It can conclude the research results just like a doctoral student turning them into an article. Of course, this combination gives it great power. You have to think of it like this: it has both a genius mind and the hands, eyes and desire to interact with its environment. Because it is ready to work forever to complete the task given to it.

As you can see, this is not an ordinary chatbot. It is a system that can do research, solve problems and even find ways to optimize itself. This is very important and directly related to DeepSeek. I will talk about this a little more in a moment. That is why people see this model as a sign of an AI explosion. It is useful to talk a little about what the AI ​​explosion or the sudden take-off of AI means. The AI ​​explosion is a concept that was put forward by AGI Good and detailed by Nick Bostrom and Leopold Ahsenbrenner. He says that when artificial intelligence gains a learning ability, the ability to improve itself, it will develop at an exponential rate. Leopold Ahsenbrenner explains this in detail in his article called Situational Awareness. He says that when artificial intelligence gains its own learning skills, it will suddenly start to grow by exponentially and will quickly reach the intelligence and skill level of an engineer or a researcher.

Here, there are many who believe that Open AI's Deep Research has brought us here. Because Open AI's Deep Research has the necessary skills for this to happen. In other words, it has the ability to improve itself, this is called recursive self-improvement. In this model, artificial intelligence discovers new information and applies this information to its own architecture. Then it starts to iterate. In other words, it starts to experiment with this information. They work with thousands of program parts, which we call agents, just like researchers and work on this new information. For example, artificial intelligence can analyze its own code and double its speed. An experiment was conducted on Deep Seek last week about this and they only gave the artificial intelligence a prompt and told it to double its speed and it did it. This gives people the impression that we have probably reached a rapid takeoff and leap in artificial intelligence. We have reached artificial intelligences that can improve themselves. So what kind of evidence is there on this subject?

Sam Altman tweeted that we were achieving 13% success in the exam that measures how close artificial intelligence is to human intelligence, and now we have reached 26.6%. We have doubled this in two days. Of course, the DeepSeek competition has an impact here. Another interesting example is that the purpose of the GPQA Diamond test is to measure how the performance of artificial intelligence is when doing research on questions that we cannot find with a simple search on Google. In the test, artificial intelligence is pitted against people who are at the doctoral level, that is, who know how to research and learn. Can you imagine, you are a doctoral student, you know how to research, you know how to learn, you know how to synthesize information, you know how to analyze, and moreover, you are a deep expert in the subject of the research. The artificial intelligence "o1 pro" performs as well as you. They claim that "o3" surpasses students in this regard. I am not the one who makes the claim, but directly the ones who developed the GPQA Diamond exam. It is unbelievable, right? We will improve ourselves, learn how to research, learn how to synthesize, use our creativity, and then an artificial intelligence will beat us. Moreover, it will be able to do this better than us week by week with an incredibly accelerating pace and a vertically upward performance.

Another interesting example, I just mentioned, is the post of someone named Matthew Burman who has an account on Twitter; He says I gave DeepSeek R1 a task, I just wrote a prompt to speed it up twice and it sped itself up twice. In other words, it directly wrote the code itself, did the iterations itself, the agents reached a new code from all these trials and we reached a DeepSeek R1 that is twice as fast as before. Another incredible aspect of the work and one of the things that impressed me the most is Derya Unutmaz, a Turkish doctor working in America. She works in the biomedical field, conducts research on human immunology and this subject. She says she is an expert in cancer immunotherapies in particular. Her posts on Twitter show this. She says she has started using Open AI's Deep Research and performs as well as us in cancer diagnosis. She gives two examples here. One of them says that the report written by Open AI's Deep Research was super high quality with a cancer diagnosis that is completely in my area of ​​expertise, I have difficulty explaining it. Another one has done another trial outside of his own area of ​​expertise and has performed very well there. In other words, Open AI's Deep Research can give results as good as a doctor at Derya Unutmaz's level in a super complex disease like cancer.

Another slightly emotional example is Felipe Millon, you can look at this with a bit more skepticism if you want. Because he also works at Open AI. But he says that my wife has cancer and the doctors had conflicting treatment and diagnosis suggestions regarding her fight against that cancer. We asked Deep Research about artificial intelligence and the suggestion they received from there put my mind at ease, he says and explains in a long thread how Deep Research achieved good results. Yes, the situation is developing very quickly. It is also certain that DeepSeek, the world of artificial intelligence, has accelerated this process. Because in the article he wrote, he presented extensive ideas about this self-learning artificial intelligence and I am sure that all artificial intelligence teams are currently benefiting from those ideas or they already had those ideas in mind, which accelerated the process. I can't figure it out anymore and the resources flowing here are starting to increase rapidly. Because can you imagine, even now, if there is an artificial intelligence as skilled as Derya Unutmaz in cancer diagnosis, who knows where it will go in the future. For example, I think the field of medicine is already one of the areas where artificial intelligence will attack the fastest.

There is one more issue; Sam Altman, CEO of Open AI, while introducing Deep Research, said that the skills that Deep Research has reached have reached a level where it can solve a single-digit portion of all the mental activities that humanity produces with economic value. In other words, I am putting 9%. Can you imagine, my artificial intelligence model solves 8, 9, 7% of all the mental activities that humanity creates value and has economic value, whatever it is, he says, and if we know the competition in this field, I am sure that much faster, much newer solutions will come in the coming days. In fact, Sam Altman said while introducing Deep Research that we have an even bigger surprise for you, this is nothing compared to that. Of course, I don't know, he may be a little excited. As you know, he is trying to increase the value of the company. Because he is constantly looking for investment, but there is an indisputable fact. It is going very fast and maybe we, as humanity, need to prepare very well for our final exam. What does this mean, what career opportunities are still available in such a world, what work will be done by us humans in such a world, how we should cooperate with artificial intelligence are issues that need to be discussed a lot. If you are raising children, what should be the education of the child in such a world are really issues that need to be discussed a lot. I think no one is fully prepared for these processes and we are incredibly accelerated. Get ready for the sharp vertical takeoff of artificial intelligence. I hope it has caught your attention.