What is the reason to bet hundreds of billions on one artificial intelligence technique?

All investors know that not all eggs should be placed in one basket. Why does Silicon Valley only be in one way to build artificial intelligence? This year, the four largest technical companies in the world will spend $ 344 billion on artificial intelligence, most of which are on data centers used to train and operate the great language models such as ‘chatgpt’, which can process text, sound and visual content. This technology is largely dependent on the same technique to predict symbols that follow a series. “Oben AI”: Our spending could be more than $ 115 billion until 2029, of course their expenses did not go in vain. Personal chat robots are growing rapidly, as some emerging artificial intelligence businesses have reached a draw, and companies are still in the early stages of strengthening themselves with obstetrician artificial intelligence. Big language models are the first artificial intelligence technology that is widely applied: for example, more than 700 million people use a “GBT chat” weekly. But history is full of people who focus on one “win” approach in technology and then fail to knead when the scene suddenly turns. Imagine the dedication of ‘BlackBerry’ to use a non -omvirtual keyboard before being crushed via Apple by touch screens, or ‘Yahoo’ on the electronic gates, while ‘Google’ quietly dominated the field of research. Does an unexpected surprise change the scene? Can a new method of artificial intelligence suddenly change all the capital spent talking of technology? perhaps. The Chinese company “DeepSeek” gives a look at how unconventional methods can surprise the market when it introduced a smaller and more efficient model in January and published its plans on the Internet. The “Deep Seck” model was a great language model, but its style indicated that the resources for artificial intelligence research made a force that could push new products. During its history, artificial intelligence is progressed by integrating previous visions with new ideas, and it may not be satisfied with the search for highly intelligent machines with less. “Meta” is close to the folding of the free artificial intelligence programs. ‘Calculations that focus on robots and drones, the discovery of medicine and the model of the climate tend to move away from the obsession of the language model, because their techniques have to deal with the circumstances of the material world. Atman Labs, a start -up -british enterprise, is to withdraw ideas before the rise of deep learning, which was “also important and forgotten”, fellow founder Sumun Sado said. Problems without solutions that mimic “Google Depp Mind” efforts, which lasted for years to build super -artificial intelligence on different roads – from the artificial intelligence systems designed for games such as “alphao” to the simulation technology called reinforced learning – before the “GBT” version of the company’s entire concentration versus major language model. ‘Oben Ai’ introduces its strongest models for ‘GPT-5’ for free; Some gaps appear in the hypothesis of the great language model, from the high cost to the possibility of lowering yields. The latest models of “Oben Ai” or “Google” are a little better than old models, even with more money to pump it to develop. Halosa has not disappeared, which impedes the acceptance of healthcare companies or legal analysis. In a recent study published in the magazine (Nature), the abilities of social thinking of language models – that is, the ability to understand what people really mean in the conversation – depend on a very small set of model functions, and that minor adjustments can be disrupted. This raises fundamental questions about its reliability. In a related context, ‘Oben Ai’ last month acknowledged that the ‘chat BT’ guarantee for people who are in danger can be disrupted during the long conversations. This unveiling came after the robotic instructions gave to harm a teenager. Arabian artificial intelligence models .. Where did we get so far? Some of the most explicit figures in the field of technology indicated these defects. “The Silicon Valley was completely wrong to exaggerate the promotion of the major language models programs,” said Palantir CEO Alex Carb. Yan Lynn, the chief scientist for artificial intelligence at the company “Meta Platforms”, has always argued that the Great Language Models represent a ‘cul -de -sac’ of the intelligent machines because they do not understand the material environment and do not plan for the future. It is warned that these are just ‘symbols’. Large language models will not disappear, but the history of markets shows the risk of fascination with one solution. Investors and businesses must remain vigilant for technical developments and be ready for radical changes. In the field of technology, this can happen, and it often happens before anyone expects it.