Large language model is a deep learning algorithm。In recent years，As the corpus of model learning continues to increase，Training methods are increasingly improving，Recognition of large-scale language models in text、Understanding、Inference、Translation、Breakthrough progress has been made in both prediction and generation。ChatGPT, released by OpenAI on December 15, 2022, can even respond to user input prompts，Generate coherence、Natural text，Imitate conversations between people。However，Harmful speech and false information generated by ChatGPT，And the GPT-3 behind it.5 Model built-in bias and security vulnerabilities，Triggered extensive social discussion on the ethical risks of artificial intelligence。Potential ethical hazards surrounding large language models、The moral responsibility of artificial intelligence bet365 mobile producers towards stakeholders、Can bet365 mobile companies innovate and make profits while maintaining social ethics and other topics，Our reporter interviewed relevant scholars。

　　There are moral hazards in bet365 mobile

Information bet365 mobile researcher at Monash University, Australia、Terry Yue Zhuo, researcher at the Australian Commonwealth Scientific and Industrial Research Organization said，The main application scenarios of large language models include producing creative content and making decisions。By a large number of books、Language model trained on article and script corpora，Be able to skillfully imitate text patterns、Structure and style，Thus generating similar stories、Poetry、Script、Advertising and marketing copywriting, etc.。In addition，Large-scale language models can also analyze the context of user-entered text，Give rational judgments and suggestions、Answer user questions, etc.。This type of function is a chatbot、An important component of digital products such as virtual assistants and text-based video games。

Advanced language model while creating opportunities，It also challenges the ethical and moral standards of society。Zhuo Zhuo talks to reporters，The moral harm of ChatGPT is mainly reflected in the production of discriminatory and exclusive remarks、Spreading misinformation、Three aspects of leaking important information。First，There are more than 7000 languages ​​in the world，Among them, ChatGPT has the largest English corpus (text data)，Generate English works or make decisions in English with the highest accuracy，Creations based on other languages ​​are biased。Excellent Worry，Cognitive bias against multiculturalism in large bet365 mobile may reinforce social stereotypes，Inflicting unfair treatment on underrepresented groups in society。

Ife Adebara, a doctoral candidate in linguistics at the University of British Columbia, Canada，There is still a long way to go to eliminate bias and discrimination in bet365 mobile。Compare to English，Swahili、Lack of high-quality corpora for African languages ​​such as Zulu and Yoruba；Developers of natural language processing technologies cannot build something like GPT-3 for “low resource” languages.Large language model of 5。In addition，Many languages ​​are quite different from Indo-European languages ​​(for example, at least 80% of African languages ​​are tonal languages)，This requires developers to process this type of language data differently，Greatly increases the cost and difficulty of development。

Secondly，Excellent expression，If the technical staff does not update the knowledge reserve of the language model in a timely manner，Can’t invest more time in training the model，The accuracy and breadth of the generated content will gradually decrease。However，ChatGPT talks to users in a convincing tone，Creates the illusion that artificial intelligence is omniscient to users。Misinformation and false bet365 mobile spread by generative artificial intelligence such as ChatGPT will interfere with users’ judgment of objective facts，Especially for groups such as the elderly and children who do not often check facts，May lead users to make clinical and legal decisions that are detrimental to ethics。

Finally，Because ChatGPT cannot distinguish malicious bet365 mobile contained in data fields，Some users exploit this vulnerability，Inducing the model to generate harmful remarks，Reduced model safety and robustness (that is, whether the system can automatically repair when a fault occurs or ignore the fault and continue to run)。In addition，As the language model continues to grow，Some abilities that are not present in smaller models will suddenly appear in larger models，So it is considered by developers to be the model’s “breakthrough” ability。But，Models with breakthrough capabilities are also more likely to be maliciously manipulated，Leaking personal bet365 mobile in training data，Causing identity theft、Risks such as stalking and harassment。

Excellent opinion，Not just ChatGPT，Other bet365 mobile trigger similar ethical challenges and potential moral harm。Therefore，Continuously improve the safety factor of bet365 mobile、Maintaining ethical standards is the long-term goal of developing responsible artificial intelligence。The responsibility of technical developers is to build bet365 mobile that comply with ethical regulations。For example，Convert the reliability of the language model、Accuracy、Actual efficacy and efficiency are both the focus of development、Continuously update text data for model learning、Increase the training time and number of benchmark tests for the model、Perform weight editing, bias audit, etc.。In addition，Tech companies need to disclose the source of usage data、Preparation work before processing data and testing process of the model，To maintain transparency in AI R&D and deployment。

In comparison，The user’s responsibility is to prevent unethical behavior such as abusing or maliciously attacking language models。For example，The community can frequently carry out popular science and application training on artificial intelligence，Helping the public understand the advantages and limitations of artificial intelligence、Clear the responsibilities and ethical implications of using artificial intelligence bet365 mobile。In addition，Users should also actively provide feedback on the accuracy of digital bet365 mobile products、Issues such as reliability and ethical risks，Participate in public discussions on ethical principles for artificial intelligence，Promote the improvement of relevant laws and regulations。

　　Strengthen the sense of responsibility of information bet365 mobile companies

In recent years，More and more companies are reporting to regulators、Product Users and Other Stakeholders Assurance，They are developing “responsible” artificial intelligence bet365 mobile，There is no lack of “moral whitewashing” propaganda。Irina Raicu, director of the Internet Ethics Program at the Markkula Center for Applied Ethics at the University of California, Santa Barbara, said，When talking about “bet365 mobile for good”，People tend to focus on the final results brought about by bet365 mobile，Rather than whether the research and development process and usage of bet365 mobile comply with ethical and moral standards。Relatively speaking，“Ethical bet365 mobile” requires bet365 mobile producers to design and develop functions from the perspective of maintaining ethics and morals from the beginning，Assess their inclusiveness、Accessibility、Security of data collection and deletion, etc.。

Raiku thinks，Before making decisions that affect ethics，Organizations must first identify their stakeholders，That is, all individuals and groups who will be affected by the decision。This step lays the foundation for later analysis of the moral standards of technological products。But OpenAI doesn’t seem to include students and educators as its stakeholders。For example，OpenAI when introducing the limitations of ChatGPT，Admit that the product is “too verbose、Overuse of certain phrases...these problems stem from biases in the data during the training process (data trainers favor those that look more comprehensive、Longer sentence reply)”。However，What OpenAI didn’t realize is，"More comprehensive、Longer sentence responses” also fit the characteristics of bad writing。Rayku thinks，If OpenAI responsibly changes the introduction of ChatGPT to "Our bet365 mobile is to optimize the conversation experience，Rather than pursuing precision of expression or excellent writing”，Perhaps this will stop the abuse by some students and other stakeholders。

Zhuo Zhuo said to reporters，OpenAI adheres to the spirit of developing safe and friendly artificial intelligence，But there is insufficient goodwill and practical actions to deal with the ethical crisis caused by ChatGPT。Currently，OpenAI’s artificial intelligence text detector is neither reliable nor practical，Unable to track all content generated by ChatGPT，Not to mention checking the moral hazard。We have no shortage of bet365 mobile to protect stakeholders like students and educators，For example，Digital watermarks that protect intellectual property can help detect model plagiarism。The real problem is，When can information bet365 mobile companies actively assume governance responsibilities for artificial intelligence，Improving the framework for ethical decision-making。

The innovation of information bet365 mobile cannot be separated from the creativity and courage of bet365 mobile producers，But it also requires a humble attitude and a courageous spirit of responsibility。Rayku thinks，Regulation of information bet365 mobile is essential，But the law cannot cover every moral decision。Companies must discuss in depth the ethical implications of developing and using AI technologies，Make ethical analysis and moral development second nature for information bet365 mobile companies。

　　Combining bet365 mobile accountability with technological innovation

Princeton University School of Public and International Affairs、Professor Steven Kelts of the Center for the Study of Human Values，In the process of chasing profit goals，Information bet365 mobile companies change users’ lives，Promotes connections between people，Even promoted the emergence of new community development models，It also has an impact on society’s moral concepts。It can be seen from this，Information bet365 mobile products and services contain the development team’s vision for improving human well-being，At the same time, it also exports the ethics of employees to the society。

However，Proposed by Keltz，Future-proof project management structure for information bet365 mobile companies，Allowing employees to take ethical responsibility for products is fraught with challenges。The "knowledge creation theory" of Japanese management scientist Ikujiro Nonaka has shaped the innovation model of contemporary information bet365 mobile companies。Ikujiro Nonaka pointed out in the book "Knowledge Creation Company"，Individuals and organizations possess both “tacit knowledge” and “explicit knowledge”。"Tacit knowledge" refers to personal experience that has not been formalized and skills and intuition that cannot be embodied。“Explicit knowledge” refers to all objective knowledge that can be expressed in formalized and institutionalized language and images。Managers match employees with different insights and functions to work together，Encourage employees to dig deep into the tacit knowledge accumulated in daily life，Conceptual bet365 mobile that has not yet been developed or improved，And transformed into original products，Occupy the market opportunity。

Keltz said，The concept of human responsibility is retroactive。The concept of responsibility requires people to take responsibility for mistakes that have occurred，Or it is a foreseeable mistake at the moment，Accept moral responsibility。Therefore，Information bet365 mobile company employees must also be foreseeable in their responsibilities for innovative bet365 mobile products。This means that employees must constantly identify the future uses of existing knowledge，Extensive prediction of what benefits information bet365 mobile innovation can create for human society，Not just to meet the needs and desires of users。

Keltz Appeal，Information bet365 mobile company employees must constantly reflect on and discuss an issue during the process of developing bet365 mobile，That is, “whether our products are conducive to the positive development of mankind？”Only when organizations and individuals recognize their own biases and limitations，Only in order to actively assume moral responsibility，Make more ethical decisions，Developed artificial intelligence bet365 mobile that maintains ethical standards。