David Sachs, the Tsar Ramsar and the US governme’s artificial ielligence, said in an ierview that artificial ielligence will be one million better in the next four years. He has put forward his prediction based on the progressive progress of three differe areas: language models, chips and calculations.
David Sax said in an ierview that artificial ielligence will be one million more in the next four years. In his view, the rate of progress of this technology in at least 3 key dimensions is expone. Expanding or exponeial growth increases with a coinuous song.
David Sax’s commes on the evolution of artificial ielligence in the future
David Sachs attributes one million to artificial ielligence to improve in three areas; The first is the models of artificial ielligence. According to him, the models have improved with a rate of 3 to 4 times a year. These models are just faster or better, and their progress is quite evide. He says we first started with chattots of the usual large language (LLM); Then we we to reasoning models and have not even reached the ages (ages), but this will be the next big leap after the reasoning models.
The next case is the progressive progress of artificial ielligence. The US governme’s artificial ielligence Tsar says each generation of chips is probably 3 to 4 times better than the previous generation. He says it is not only every single chip that get better, but differe companies discover new ways to network them.
The third, which affects the evolution of artificial ielligence, is computing and processing. “Just look at the number of graphics processors (GPU) deployed in data ceers,” he says. When Ilan (Mask) first started teaching Grak, I think they might have 100,000 graphics processors. Now they have reached 300,000 and are on the way to one million. The same is true of the Openai data ceer, and may reach 5 million graphics processors or 10 million graphics processors within a few years. “
How will they evolve one million equality by multiplying the progress of these 3 artificial ielligence factors? If in the first two years of the progress of models, chips and calculations each have grown by 10 times, it will be 100 times higher in exponeial growth in 4 years. If we get 100 to 3, we will reach one million.
