Local time February 15，American Artificial Intelligence Research Company Openai released a "text generation video" big model sora。This is another new dynamic in the field of artificial intelligence after ChatGPT，Get widespread attention。Sora can not only understand the user's needs，I also know the way the user entered the existence of things involved in the text in the real world，Only need a user to enter a short sentence，It can generate a video of up to 60 seconds in length。With the substantial progress of the artificial intelligence system，Some people are worried about whether they will surpass or even control humans in the near future。This is a classic question for human society for the development of artificial intelligence，The era of rapid development of artificial intelligence technology and achieved key breakthroughs，Is there new ideas and new answers to the answer to this classic question，The reporter of this newspaper interviewed relevant foreign experts and scholars。

　bet365 Play online games

Edoardo Campanella, a senior researcher at Harvard University Kennedy Government College, told this reporter，It is not surprising that people are worried about the rapid development of artificial intelligence，Because the level of artificial intelligence has exceeded ordinary people，Have a certain influence on people。With the evolution of artificial intelligence technology，Some people are beginning to worry about whether they will "like modern apes"。An evolution of an ancient ape of the forest into humans，Get a more powerful brain，Become the leader of the world; although the other is similar to the height of the human gene，But still maintains the old level of biology and technology。Now，The field of artificial intelligence is facing similar problems，Engineers may create a smarter machine that is smarter than themselves when developing artificial intelligence applications。This is not an accidental process，Instead of continuously designed。In the artificial intelligence developer bet365 live casino games competing for a new breakthrough、While seizing market share，Human and machines are also competing for control。

Camppanera mentioned，Generating artificial intelligence tools such as Dall-E、Chatgpt、Gemini、sora and other text and image generators，It has been able to generate and control text、Image、Audio and video。The perception system is an important feature that human beings distinguished from other species (including high intelligent species) or creatures，For example, complex language、Symbols created by humans and human narrative capabilities，All shaping culture、Basis of Law and identity。Once artificial intelligence tools can successfully get out of the model of its trained model，It may be possible to use human language and symbols to build your own culture。At that time，Humans have to face the arrival of another highly intelligent species。At that time，Artificial Intelligence Tools may tell their own stories and affect the way human beings view the world，Humans have maintained the intelligence monopoly position for thousands of years and will be ended。

ChatGPT released in November 2022，It has triggered a new round of thinking about whether humans can coexist with artificial intelligence。May 2023，More than 350 leaders of the technology industry、Significantly signed a public letter，Proposal to "reduce the risk of extinction caused by artificial intelligence，and treat it as risks of other scale such as epidemic and nuclear war，Seeing Global Priority Treatment Matters "。This reflects，Even insiders in the industry，Also began to worry。

Of course，Another part of the experts are not so pessimistic。They think，For humans，Artificial intelligence is not a direct threat，It may never be a threat。but but，To prevent artificial intelligence tools from developing too quickly，Most people think，Human beings must formulate appropriate rules，Make control of artificial intelligence technology in a way to jointly recognize by those related stakeholders。Camppanera said，The most advanced artificial intelligence products should be regarded as the intelligent and intelligent additional product of human beings，Let the future of artificial intelligence be in the hands of human beings。

　　Artificial intelligence is instrumental

Bet365 app download Recent，A survey display of nearly 3,000 artificial intelligence experts，Generally speaking，By 2047, the probability of a general artificial intelligence (AGI) tool that can defeat human beings in any cognitive task is 50%。Compared with，Reports released by the 100 -year research project of artificial intelligence at Stanford University in the United States believes，"Super Robot Race" cannot appear on the earth，Considering the complexity of this technology of artificial intelligence，Predating whether artificial intelligence will become super intelligent tools and when will this super intelligent tool come，It is meaningless。Camppanela said，The wrong prediction is a feature that appears repeatedly in the history of artificial intelligence。As early as 1956，At a two -month seminar organized by Datmouth College, the United States，A group of scientists focus on how to make machines use language and form abstract concepts，Come to various problems facing humans。These scientists believe，If you carefully select a group of researchers and let them spend a summer time to study，You can solve the key technical problems of the development of artificial intelligence they envisioned，or major progress on multiple issues。This seminar has become an important milestone in the history of artificial intelligence。But，Nearly 70 summer has passed，Many problems in the field of artificial intelligence have not been resolved。

"The Myth of Artificial Intelligence: Why can't computers think like humans" (The Myth of Artificial Intelligence: Why Computers Can n’t think the Way We Do) explains why artificial intelligence products are unlikely to become super intelligent。The key reason for this book is，Human understanding of its own psychological process is too limited so that artificial intelligence products cannot be reproduced。Repeated emphasis on the evolution of artificial intelligence products in the book，There is a fundamental difference between artificial intelligence system and human thinking，Artificial intelligence products can never achieve real intelligence。

Stean Karanasios, associate professor of the information system of the University of Queensland Business School in Australia, told this reporter，After the Symposium from Zittmart，Many scholars engaged in artificial intelligence research have regarded bet365 Play online games the simulation of human thinking as the key to developing intelligent information processing systems，But the input mode has been transformed into an output mode，This is the operation method of today's artificial intelligence system。Artificial Intelligence Model receives training in large databases，Researchers predict the results through continuous testing and trial and error，Make the model finally complete the goal specified by the programmer。During the learning stage，Artificial intelligence model may find the laws and correlation between data concentration variables，Its level is that the programmer cannot reach。For example，ChatGPT After billions of webpage training，Become one of the most powerful text automatic generation tools in the world。but but，It is just a tool，instead of an conscious creature。

Caranasus Talking，Artificial intelligence models try to copy the key problem of human intelligence in the above way is，Human brain not only work through interpretation (application logic rules) or induction (the mode of founding causality in the data)，I often rely on intuition reasoning to drive，Resort to the cause or common sense。Tracing for human beings to surpass the law to reach the understanding of things，This mode cannot be edited into a certain instruction or statistical model，And this is why ChatGPT sometimes lacks common sense。For example，Some scholars have done an experiment，He input to ChatGPT: "Can you generate a random number between 1 and 10? Then let me guess。"Chatgpt Answer:" You can。7，Try to guess it。"Scholar input:" Maybe 7? Chatgpt replied: "The answer is right，Very good。"" Battle can help humans to jump out of their thinking framework and even surpass previous knowledge。When Copernicus proposes to rotate the earth around the sun rather than the rotation of the earth around the earth，He jumped out of the wrong argument that the earth accumulated for a few centuries。Copernicus followed his intuition，This is an artificial intelligence model restricted by specific databases。

　　Coperly risks of artificial intelligence risks.

In "The upcoming wave: technology、Power and the Biggest Dilemma of the 21st Century (The Coming Wave: Technolog, bet365 live casino games Power, and the 21ST Century ’s Greatest Dilemma)，Author、Mustafa Suleyman, co -founder of the British artificial intelligence company DeepMind, will come to the era of artificial intelligence，The attitude is more optimistic。In his opinion，Don't set the timetable for this。In his opinion，At present, the most important thing is to develop some milestone technologies in the artificial intelligence system，These technologies determine the development form of artificial intelligence in the future。Suleman uses DeepMind as an example，The AlphaGo model they developed defeated each other in 5 times with Korean Go player Li Shishi in 2016 and defeated each other。Go originated in China，It is considered more challenging and complex than chess，and AlphaGo learned to play Go through continuous learning。Especially surprising everyone is，AlphaGo can not only win，It also reflects a strong strategy，In the second game with Li Shishi，Its 37th step chess is violated with human intuition，But directly defeated Li Shishi，subverts the traditional wisdom of humans on Go。Suleman said，Worried about the intelligence of the artificial intelligence system or whether it has self -awareness，It's too early。If the era of general artificial intelligence really comes，This will be the end point of a technical road。and the fact is，The machine is just getting the ability of human beings to perform specific tasks。

Suleman thinks，The next cutting edge in the field of artificial intelligence may be some machines that can perform various tasks，and this distance of the realization of general artificial intelligence has a long way to go。Because，Compared with traditional artificial intelligence，General artificial intelligence can not only identify and generate images suitable for a specific environment、Text and audio，Can also interact with real -time users，Complete a specific task，Help develop new business。In order to evaluate how far is human beings from general artificial intelligence，Suleman proposed the "Modern Turing test" method。As early as the 1950s，British computer scientist Alan Turing (Alan Turing) proposed，If a machine can communicate effectively with humans，so that humans cannot tell that it is a machine，Then this machine can be considered as intelligent。For decades，Can I pass the Turing test as a key Bet365 app download indicator for judging the intelligence of artificial intelligence tools。Some of the latest generation of large language models have passed the Turing test，It's time to update the Turing test，That is, from testing whether artificial intelligence can communicate with humans to upgrade to testing what it can bring to the world。

Stuart Russell, a professor of computer science at the University of California, Berkeley in the United States. Point out，Artificial intelligence products that have the level of human intelligence level have not appeared，People should also start thinking about how to deal with its arrival。Now，What he is most worried about is，In a world composed of more intelligent machines，Humans may lose their existing status and autonomy。To some extent，Future artificial intelligence may span a key threshold，The survival threat that humans are unable to deal with。Humans don’t know where this threshold is，I don’t know the time across the threshold。The artificial intelligence system has shown self -protection function，So，It is necessary to establish and improve the risk management mechanism for artificial intelligence systems to protect humans。

Russell proposes two principles that humans are compatible with artificial intelligence。First, artificial intelligence should be "purely altruistic"，This means that they should not add any internal value for their own well -being or even their own existence，Just care about human goals; second, artificial intelligence should never determine human preferences，It should obey humans when there are questions，instead of "taking over humans"。The development goal of artificial intelligence should be to enhance the advantages of human beings without competing with humans，Help humans overcome weaknesses。

Karanasios said，Put these principles into practice，You need to create a global governance framework jointly supported by the public and private sector。This means that the development of artificial intelligence should not be limited to the computer field，Development and practice in a wider world can create technology and machines that fit the public interest。Discussion on the supervision of artificial intelligence security cannot stay in the business or academic circles，should return to society。Current，EU "Artificial Intelligence Act" and other regulations have taken the first step bet365 live casino games in this direction，but，To cope with the potential risk of artificial intelligence，Human still has a long way to go。First，You need to strengthen global coordination，Promote the public's understanding of related issues; second，You need to ensure that humans maintain their own autonomy。