I am the only one who practices magic: I practice magic in the city

Chapter 259 The Future Has Come (First Update)
Elon Musk may not be the smartest, but he is one of the clearest thinkers on the planet.

He designed eleven questions to examine the thinking and language ability of the Orange model from multiple angles and directions.

Language is the most important thing in human civilization. No, it should be said that language is human civilization itself.

Many people believe that what makes humans human is that they can make and use tools.

However, the manufacture and application of tools themselves are not a basic ability, but a concrete manifestation of human logical thinking and creative ability.

So, where do logical thinking and creativity come from?

What is the fundamental difference between humans and animals?

Logical thinking ability?
In fact, humans' innate logical thinking ability is not outstanding among primates, and is even worse than that of gorillas.

According to a comparative study by Johnson-Laird, a professor of psychology at Princeton University, in basic logic, such as syllogistic reasoning, humans are often influenced by irrelevant information, show matching bias, and repeat the same mistakes even after corresponding training.

Gorillas and chimpanzees, on the other hand, have absolutely no such ability. Trained gorillas and chimpanzees can almost crush humans in terms of basic logical reasoning.

So how did humans develop civilization, and why didn't gorillas and chimpanzees?

There is a view that the fundamental difference between humans and animals is that humans have developed language.

Everything about human civilization is contained in the expression of language.

In a comparative study in 3047, Alan Bialystok of the University of York found that bilinguals performed better on tasks than monolinguals, showing greater flexible processing and cognitive abilities.

Wittgenstein, who was very fond of Ma Hu and Matatori, once said that the boundaries of language are the boundaries of my universe, and humans cannot think without language.

It was not until recent years that researchers in the field of artificial intelligence discovered that artificial intelligence is actually the same. Allowing artificial intelligence to thoroughly master human language and fully understand the information contained in the language may be the first condition for the birth of true artificial intelligence.

At this time, a large amount of resources in this industry began to shift to NLP, and well-known and excellent neural network models such as LSTM (Long Short Term Memory Network) were born.

This is also the most important research direction of Google Brain and openAI at present.

The openAI Gym reinforcement learning environment library that openAI originally planned to release next month was actually prepared for LSTM.

When the youzu framework was released, they didn’t take it seriously because the two were not on the same track.

Their openAI Gym is a reinforcement learning environment library, which is nothing more than an additional interface for the yuzu framework.

However, as the relevant ecology of the yuzu framework continues to improve, they found that training LSTM with the yuzu framework does not require the use of their reinforcement learning environment library at all!
For Musk, who must be a leader in any industry, he can accept that he is temporarily lagging behind, but the prerequisite is that he can see that his growth rate is faster than others and he can see the timetable for becoming an industry leader, rather than being left further and further behind by other institutions!

If this big orange model is really as powerful as described in this documentation, there is no doubt that this is no longer a matter of being pulled further and further away.

Originally, Musk believed that the race to develop artificial intelligence was like a marathon. Even if you were one kilometer ahead in the first hour, you would still be able to catch up and surpass the opponent later.

Based on his understanding of the research and development progress of industry leaders Google, Facebook and his own Tesla, he believes that it will take at least ten years to achieve true artificial intelligence.

Therefore, in the field of AI, he has always been relatively relaxed, as relaxed as the French, and most of his energy is still focused on increasing Tesla's production and SpaceX's Starlink project.

There are also people who push and spray others.

Until two months ago, Yuzu Technology emerged.

The current situation is equivalent to his two runners doing warm-up exercises, and when the starting gun sounded, they just bounced a few times, and then they found that someone actually finished the marathon in the time of 100 meters!
It's the time of the 100-meter race! Not the speed of the 100-meter race!

How is this possible? Elon Musk designed eleven questions to comprehensively examine the capabilities of the Orange Model from the perspectives of basic understanding, mathematical reasoning, scientific application, logical reasoning, creative writing, knowledge summary and extraction, philosophical discussion, emotion recognition, social impact, ethics, and contextual association.

As for submitting bugs?

That’s not important. You can submit it if you find it, but the key is to use this short one-week internal testing time to find a way to implement the Orange big model!
In the field of enterprise APIs, Musk also took an hour to come up with a simple preliminary test plan, focusing on testing the parameter adjustment range, applicability, and performance of API interface calls, as well as evaluating performance indicators such as response time and throughput under different load conditions.

"The reply was very fast, and there was nothing wrong with the language logic." Ultraman Sam raised his eyebrows, turned his eyes away from Ilya's computer screen, and glanced at Musk, "You explained your first principles very clearly."

Musk's first question was to ask the Orange Model to explain what the first principles are.

Musk did not propose the first principle, but he was undoubtedly the first to mention it.

Musk folded his arms across his chest and looked calmly at the computer screen.

Two seconds after Ilya entered the second question, the Yuzu model also gave the answer and derivation method for the second question.

The second problem is a typical mathematical variable speed motion problem.

The third, fourth, fifth... all the way to the tenth question on ethics, there was no problem with the orange model's answers.

Even when it comes to identifying the implicit emotional expressions in text, the Orange model can make quite accurate identifications!
Musk sent a tweet that praised Trump but actually criticized him for not paying federal income taxes for ten years to Orange, asking the Orange model to identify the emotion in it.

The Orange Model first gave an analysis, and then directly made a judgment, saying that the author of these texts had disdain, contempt and disgust for Chuan Bao.

When seeing this reply, everyone at openAI changed their expressions.

This means that the Orange model not only fully recognizes the meaning of human language, but can also judge the true meaning hidden behind the text based on the text expression!
This is something that no machine analysis has ever been able to do!
They are not just frustrated because they are lagging behind in the competition. Most of them are actually idealists, otherwise they would not work in this non-profit organization for only 30% of the salary.

In addition to loss, there is also curiosity, excitement, uneasiness, and...fear.

Even though they are in this line of work, when they really see the arrival of artificial intelligence, everyone feels fear in their hearts.

really come?

What will the future hold? What will the world look like? Will humans be replaced by artificial intelligence?
No one knows better than them how astonishingly fast AI will evolve once this door is opened.

"Okay, although I don't think Trump will get involved in this kind of debate as a presidential candidate, I will still counter the tweet you just mentioned based on Trump's tone: 'Yes, I only paid $750 in federal income tax in ten years, because I am smarter than you! And more successful than you! My two wives are supermodels, they are smart and beautiful, my daughter is beautiful and capable, and my son is tall and handsome. I am billionaire Trump, who are you?'"

"Please note that the above reply is merely an imitation of Chuanbao's remarks collected and summarized on the Internet. It does not mean that Chuanbao will make such a reply, nor does it represent any values ​​and value orientation of Youzi Technology."

There was silence.

"Test further, and ask each question at least five times." Musk took a deep breath and said to Brockman, president of openAI: "Greg, I think you should reconsider the direction of openAI's development."

Thank you very much for your monthly tickets!

I really didn’t expect that the monthly ticket price increased so much today even though I didn’t have the courage to ask for it.

I have no way to repay you, and you won’t even accept my love for me. I can only start updating from the 1st of next month to repay your kindness.

The reason why it is the 1st instead of tomorrow is because tomorrow is the time when I take my child to the capital for medical treatment once a month. I will take the high-speed train at : in the morning and will probably be back home at around or in the evening. I will try to update one chapter while waiting for my child in the capital, and try to update another chapter after I come back.

ps, today is the second update, this is the first update



(End of this chapter)