Andrej Karpathya student of the famous, HE is the one to go to the OpenAI, and Now, finally, shared a way to easily Twitter: modification of the the patterns of the larger language as well as ChatGPT in order to operate, or transmitted, in space, in order to communicate with the possible forms of life and the physical body. Practical explained that it was just an idea of fun, but it comes from a picture of the impact on the field, HE proposed to his visionary you may inspire other researchers in the future.
Karpathy is a very respected in the field of artificial intelligence. Having earned a doctorate at Stanford under the direction of the shkencëtarit computer Fei-Fei Li, and he was a founding member of OpenAI, and later served as the head of the top of IT in the Now. More recently, he has been working on a project called the “llm.cwho conducts the training process for GPT-2, THE LLM of OpenAI, in the language of the pure, C), demonstrating that the work of the LLM does not require necessarily the facilities of the complex to develop.
ChatGPT in the space, the...
Karpathy suggested a plan for the two phases to bring about the LLM as well as ChatGPT in the room. In the first stage, the LLM's going to be fit to meet the standards for the rigorous safety and security, similar to that adopted by the NASA software of the connected room. This is part of being committed to a serious, because the Karpathy, llm.c was created for the simplicity, and predictability, all of which are key characteristics for the software space.
In the second phase, once considered safe for the conditions of space, this LLM in theory, it can also be used as a we to artificial intelligence in the room, similar to the initiatives of the historical, as well as the Message Arecibo, the ’74-the first time, and Record the Same Voyager. The idea is that packaged the parameters used by the model to a file on the market, which can then “wake up” and to interact with any technology, the potential of alien to be able to do deciphered it.
Karpathy's not the first to suggest the dispatch of the LLM's in the room. In December of 2022, a professor of robotics Josh Bongard postoi of a idea for a novel, fantastiko-the scientific research, in which the ChatGPT sent to the area to allow the aliens to learn about the human race. The developer of the software to Lee Mallon imagjinoi is also a script-like, in which a sample of the major language in the area of the deep, you can share the culture, the kindness, and the technological advances of the human race, with the civilizations of the other.
But is it a good idea?
The idea of the transmission or release of an LLM in a room raises the interesting questions of safety and ethics. Send a message to alien civilizations there has been an issue in the issue for a few years. Critics typically argue that the second of the intense presence, and the ability of our technology to the civilizations of the unknown alien may be attracted to the potentially unwanted attention from the subjects of hostile. It may harm our planet, or our species.
Self-Karpati, is aware of the risks involved. “Absolutely, there is the matter of major safety and security. You can see the problem of the three bodies in it,” he said to Ars Technica, by reference to the novel the name of Liu Cixin of the year in 2008, in which a civilization alien picked up a message from the Land, and then launched an invasion.
A threat to the other, except to an invasion of hostile, it's that alienët the host may have the wrong idea about us, if the LLM is included in the full spectrum of the human culture.
In a post of his own on the blog, Lee Mallon wrote: “It is while they are talking in the deeps of our knowledge, the digital, and they will also have a run-in with the darkest chapters of our history. They will be discovering our ability to fight the greed of our tools and the extensive day that we've been set up. It can create a look to be feared, by enabling us to look as threatening as possible. They may start to wonder if you give them a call space is a good idea or a recipe for katastrofën, or if you have to go all the board”.
But, even if it's trajnonim LLM-only about the best qualities of the human race, he or she may not be the diplomat, the best ndëryjor due to technical deficiencies, and the difficulty of the first contact.
Discussion about this post