Shulou(Shulou.com)11/24 Report--

Thanks to CTOnews.com netizens for the clue delivery of the holy Buddha! This article comes from the official account of Wechat: SF Chinese (ID:kexuejiaodian), author: SF

Friends who have seen the movie Matrix may remember that in the film, human beings live in a world controlled by artificial intelligence, which is clearly arranged by AI from birth to death. What the hero Neo and other awakened humans have to do is overthrow the rule of AI. AI acquires consciousness, which in turn threatens human beings, which is a common passage in science fiction. In real life, many people worry that AI will have consciousness.

(by Mei Lin / tr. by Robert Taylor)

At the beginning of this year, the emergence of ChatGPT exploded the scientific community and public opinion, so that the topic of AI, which people thought was more science fiction, really came into our lives.

In May, China launched the standardization of large models. On August 31, the large model products of eight enterprises and institutions, including Baidu, Byte Jump and the Chinese Academy of Sciences, became the first batch of AI models to be filed through the interim measures for the Administration of generative artificial Intelligence Services, which can be officially launched to provide services to the public.

alert! Conscious AI will threaten the development of human AI, which brings us both opportunities and challenges. For a time, there was a wide debate about AI in the public opinion field.

Those who support "AI consciousness threatens mankind" are pessimistic about this. They are warning that AI tools such as GPT (Converter-based generative pre-training Model, generative pre-trained transforme) may pose a threat to humans. Some of them believe that GPT can be the first "spark" to ignite the emergence of universal human intelligence (artifificial general intelligence), while general artificial intelligence is a perceptual, conscious machine whose intelligence can surpass that of humans. Compared with them, we humans will become insignificant.

Opponents argue that there is no scientific basis to prove that large language models now have or will have the ability of subjective experience and perception, which are only possessed by humans or animals with higher intelligence. In other words, AI pessimists seem alarmist. Technical optimists feel that AI does not have our life characteristics at all-we perceive the world around us through our senses to form consciousness, while AI is just a machine and cannot interact with the world in a similar way, so it does not form consciousness at all.

As the "father of artificial intelligence" and French computer scientist Yann LeCun said, today's AI is unlikely to develop into a general-purpose AI, so they do not pose a serious threat to human beings. However, these optimists are also overly optimistic-the threat posed by AI to mankind is real.

Today's AI, even if it doesn't have consciousness, makes some people give up some of the abilities that AI is not good at, but some of which are unique to human beings, such as writing novels. Now, some AI have been trained to help human beings write novels. As a result, some authors hand over the work of creating plot progression to AI, so that they can be freed from the creative work of wasting their brains. In the long run, the "thinking" that mankind is proud of can be completely replaced by AI, and we will be less and less capable of thinking.

We personally tested the outline of the novel generated by "Wen Xin Yi Yan". Both sides have their own reasons: pessimists' worries are not unreasonable. Comparing how AI handles data and information and making decisions is still a "black box" for us, and we are not sure whether AI will have consciousness in the future. The views of optimists are more practical, pointing out the essential differences between AI and our human beings, as well as the problems in our dependence on AI at this stage.

As to whether AI will become the "master" of our human beings, as described in the Matrix, further research is needed by scientists.

How to judge AI? By planning ahead, some scientists are already at the forefront of dealing with conscious AI. Recently, a team of 19 computer scientists, neuroscientists and philosophers published an article (not peer-reviewed) pointing out that according to some theories of human consciousness, they have proposed 14 criteria for judging whether artificial intelligence has consciousness.

These are the 14 standards listed in the paper. The more criteria AI matches, the more likely they are to be conscious. (P. Butlin et al.,2023) to judge whether AI has consciousness or not, we must first define consciousness. The scientists decided to use the "phenomenal consciousness" proposed by New York University psychologist Ned Block to define consciousness. In short, consciousness is the experience of subjective feelings, such as seeing red flowers or feeling pain.

So how to explore the consciousness of AI? Unlike humans, humans can detect brain signals through EEG or magnetic resonance imaging to explore human consciousness. Scientists can't do magnetic resonance imaging of AI, can they? Instead, scientists choose to start from the theory of human consciousness, explore some of the most basic standards of consciousness, and then look for characteristics that meet these criteria in the underlying architecture of AI, in order to determine whether AI has consciousness.

Using these standards, scientists have tested some mainstream AI tools, such as ChatGPT and PaLM-E. They found that currently, no AI can meet multiple criteria at the same time, so none of them are conscious, at least for now.

References:

Https://doi.org/10.48550/arXiv.2308.08708

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.