A sobering assessment of the explosion of artificial intelligence: the military must be isolated on bases

Artificial intelligence systems have developed very rapidly in recent years.

Former CEO of Google Eric Schmidt It believes that the capabilities of apps will grow to significantly change the world in the future. Development takes place in fast cycles, meaning a new artificial intelligence model is completed every 12-18 months.

The biggest changes are related to the expansion of the so-called context window. That way, you can re-enter the answer provided by the computer to the follow-up questions.

– Eric Schmidt says that within five years, we could develop thousand-step answers that could solve critical problems in medicine, materials science, and climate change. NovemberIn an interview to the media.

There are “artificial intelligence agents”, i.e. large language models, capable of learning new things. The agent can absorb all existing knowledge about chemistry, develop hypotheses, and test them in the laboratory to increase his knowledge.

– These agents are getting stronger, says Schmidt, and there could be millions of them.

AI agents can develop their own language to communicate with each other. After that people cannot monitor their activities.

– What should we do? Close them? Do you really unplug the machine? This is really a problem when agents start interacting with each other and doing things in ways that we humans don’t understand. I think this is the limit that can be met, says Eric Schmidt.

In the West, large artificial intelligence systems are being developed by OpenAI, Microsoft, Google and Anthropic. Along with these are countless smaller players. Industry players are limited by their dependence on expensive equipment. Because of export restrictions, China is believed to be years behind.

See also  Symptoms of Whooping Cough: Cough is a symptom of the disease

Eric Schmidt’s concern is that the development of artificial intelligence is restricted in the West, but in China or elsewhere in the world, the activity could be dangerous. In addition to deep fakes, artificial intelligence may be used in the future, for example, in the development of biological weapons or for powerful cyber attacks.

According to him, states should adopt certain policies related to the development of artificial intelligence. He likens the situation to ballistic missiles, whose test launches are announced in advance to avoid catastrophic misunderstandings.

– At some point, both the US and China may have very small amounts of powerful computers with independent innovation capabilities. Eric Schmidt says the capabilities are more than we want to give to our citizens or our competitors without permission.

According to Schmidt, such computers would be housed at armed forces bases. Their energy source may be nuclear reactors. The facilities are protected by barbed wire and machine guns.

– I think it is fair to assume that there will be quite a few such systems. Other than that, there are many that are less effective and more widely available, says Schmidt.

Current and developing language models are based on the simple principle of predicting the next possible word.

– We want to avoid a situation where a reckless agent in China grabs a gun and fires it thinking it’s some sort of game. Remember that these systems are not human and don’t necessarily understand the consequences of their actions, says Google’s former CEO.

See also  Hamas continues to demand a permanent ceasefire - talks will resume on Sunday | Foreign countries

Leave a Reply

Your email address will not be published. Required fields are marked *