The Chinese tech industry recently gave the American Tech Industry – as well as the stock market – a badly shocked when it unveiled Dipic, a model of artificial intelligence that performance equal to America’s best works. Demonstrate, but it is developed in a part of it. Despite cost and trade restrictions on AI chips.
Since then, many attempts have been made to know how they did it and whether it was above the board. These are not the most important questions, and the excessive focus on them is an example of how we were trapped in the first place.
The real lesson of DiPsic is that the US point of view of AI’s safety and rules – these concerns were largely nonsense by both Biden and the Trump administration as well as many AI companies. It was never possible to overcome the spread of this powerful emerging technology, and certainly not only imposing trade restrictions on components like graphic chips. It was a self -serving fiction, which was tied to the leaders outside the touch by an industry, which the government wanted to kneel down its rivals.
Instead of useless efforts to maintain the bottle, the government and the industry should prepare our society for the sooner changes.
The misguided focus on the containment is a nasty echo of the nuclear era, when the United States and others restrict enriched uranium access, limiting the spread of nuclear bombs, keeping an eye on what some scientists are doing. By sending inspectors to the labs and sending them to the labs and restricting the spread of atom bombs, the military bases had a clear impact on the steps that are occasionally in collaboration with the show of force. The world is not yet flying.
However, an important difference is that nuclear weapons were only developed by a few scientists on the largest edge of their fields. On the other hand, the main idea of ​​strengthening the revolution of artificial intelligence has been Since the 1940s. The first thing that opened the flood route was the arrival of a wide data set (through the Internet and other digital technologies) and then the arrival of powerful graphic processors (such as NVIDIA), who could count the AI ​​model from these data trousers. Is
Another difference: Every nuclear weapon has to be made out of steel and fastened material. On the other hand, some AI models can fit on the USB stick, and only this stick can be copied and made by the plugin in the new laptop.
Initially developing a new model, such as Chat GPT, is a very expensive process, but it’s output, known as the Lord Model weightWhich are very valuable, and copyable. Companies like Open, who have strongly declared that AI pose an existential threat to humanity, kept the weight of these models with itself, lest others create even more powerful something. Paguy should be backed on these expensive development works.
And if these security -minded companies made a lot of money because of the US government’s defense measures? Well, the only price to keep humanity safe, okay?
President who was an ally of these companies in Biden – especially, said Deputy Chief of Staff for Policy, Bruce RedAfter seeing the “Mission: impossible – impossible – dead part One”, a story of Ai Gone Rogue. After already signing an executive order banning the sale of these key chips to China, Biden signed another for the establishment of safety and safety mandate.
The Trump administration is operating under the same poor logic. Only one day of his new term, President and Open’s chief executive, Sam Altman (refreshes his $ 1 million pact with Trump’s opening fund) announced a widespread computing infrastructure project. Called Star Gate, it is billed as a multi -billion dollar bid to maintain us in the fast growing industry.
Deepisic chose the next day as the moment Posted An dissertation in which the world has been allowed to go to its great uprising. I think at least they are having fun. The company says it spent only a small portion of what Open and others spent, as it was able to improve its software and train its model more efficiently. Such developments have allowed many other technologies to be available cheaper and widely available. Nevertheless, not everyone thinks that this account, especially questions has been given questions about China’s respect for intellectual property rights and trade restrictions. Could the company submit? A forbidden stash of nvidia chips? May be. Can the model’s growth cost more than disclosure? Something happens by some estimates. Openi says DiPsic May Some of his works have been stolen. I am a gut for a company that has made a commercial product as a huge part of the Internet then claimed that it is “Fair use. (The New York Times has sued Openi and Microsoft to see if the use of news content is used in their AI system.)
But whatever DiPsic did, it and others can continue to do so. Already, many AI companies are building on the model of Deep Sak. People are downloading it or questioning it for only a small part of the open AI charges.
Inside the industry, there is a popular trip that the original turning Agi, or artificial general intelligence will develop, when AI reaches human level intelligence and potentially independent. This means that what is happening now is just a kind of warm -up, which no one needs to worry about. This is an easy lie. We’ve arrived at the other AGI turning point: Artificial Ende Engeneous-AI, which is useful for sharp, cheap, expansion and broader purposes-and we need to engage with what is happening.
Many observers have described this A sputanic moment. This is wrong: America cannot re -establish its dominance over the latest AI, because technology, data and its skills have already been divided into the world. The best way for this country to be a new age is to prepare its effects.
If AI’s inevitable proliferation endangled our cyber -scorerity, for example, instead of just regulating exports, the time has come Tighten our Networking Infrastructure -It will also protect it from the current threat of hacking by random agents or adverse governments. And instead of imagining how some future bullying can attack us in the future, it is time now to use the AI ​​available now to incorporate corporations and governments into their dominance, ours Destroy rights, spoil inequality. Since this technology continues, who will be left behind? Which rights will be threatened? Which institutions will need to be rebuilt and how? And what can we do so that this powerful technology that benefits the public with so many capabilities?
It is time to admit that the interests of some of the big, multinational companies are not good for people’s interests that are not proxy that they face such a memorable change.
DiPsic has to bring us here and whatever it may have done, perhaps to force the feeling for which we can thank them.
