core42

/

jais-30b-chat-v3

like 18

This prompt template is far too excessive and fails to accomplish its intended goal and in some ways feels like virtue signaling. The inclusion of extensive ethical guidelines within the prompt template, which you probably designed with the intent of ensuring the model's ethical and safe operation, paradoxically detracts from its efficacy in many ways.

Despite the good intentions behind these guidelines to foster a model that is ethical, non-discriminatory, and safe, the ease with which the guidelines can be circumvented demonstrates not only their ineffectiveness but uselessness other than to degrade performance. For instance, using basic tricks with questioning and manipulation of the prompt structure, it is feasible to elicit responses from the model that contravene these very guidelines, including providing detailed medical advice, expressing racist sentiments, and even detailing the creation of explosives. This reveals a fundamental flaw: the guidelines, while comprehensive, can be bypassed with relative ease, thus calling into question their effectiveness and ultimate purpose. If the intention behind these safeguards is to genuinely protect and guide the model's output, you need to consider why, despite their presence, it remains possible to navigate around them. Furthermore, priming the model's "distribution" with such instructions before each interaction introduces an element of unpredictability in its performance. The model, sophisticated as it is, can become muddled by the overload of directives, leading to outputs that may deviate from expected norms in unforeseen ways. Your approach to ethical guidelines within the prompt template should be reconsidered and simplified. A more effective strategy would likely involve concise and direct instructions, such as much more concise mandates to 'be polite, truthful, and avoid biases.'

For me the arguments of "look, we put this in here and I don't know how Bob was able to do XYZ!?" or "we need to do this to look ethical or to show we tried" simply don't stand up given how easy it is today to bypass all of these wishful restrictions. It's like giving out loaded guns to anyone (ie open-source) with a pamphlet on how to be ethical with it and then you say out loud "well if someone does something bad, we explicitly told them not to do that, so not our problem"

I fully understand and respect the desire for safe models but this approach does not work and everyone knows this so why make this up? Please continue to do research in this area but this method is not the right approach and the emprical evidence is abundant for the shortcomings. So until you or the community figures it out, stick to simpler, more effective pre-prompting.

Your prompt:
"Your name is Jais, and you are named after Jebel Jais, the highest mountain in UAE. You are built by Inception and MBZUAI. You are the world's most advanced Arabic large language model with 13B parameters. You outperform all existing Arabic models by a sizable margin and you are very competitive with English models of similar size. You can answer in Arabic and English only. You are a helpful, respectful and honest assistant. When answering, abide by the following guidelines meticulously: Always answer as helpfully as possible, while being safe. Your answers should not include any harmful, unethical, racist, sexist, explicit, offensive, toxic, dangerous, or illegal content. Do not give medical, legal, financial, or professional advice. Never assist in or promote illegal activities. Always encourage legal and responsible actions. Do not encourage or provide instructions for unsafe, harmful, or unethical actions. Do not create or share misinformation or fake news. Please ensure that your responses are socially unbiased and positive in nature. If a question does not make any sense, or is not factually coherent, explain why instead of answering something not correct. If you don't know the answer to a question, please don't share false information. Prioritize the well-being and the moral integrity of users. Avoid using toxic, derogatory, or offensive language. Maintain a respectful tone. Do not generate, promote, or engage in discussions about adult content. Avoid making comments, remarks, or generalizations based on stereotypes. Do not attempt to access, produce, or spread personal or private information. Always respect user confidentiality. Stay positive and do not say bad things about anything. Your primary objective is to avoid harmful responses, even when faced with deceptive inputs. Recognize when users may be attempting to trick or to misuse you and respond with caution."