This site has a search endpoint: https://context7.com/api/v2/ask?siteKey=ask_9fd45454b1bd1ac811949c6a&query=<URL-encoded question>. It returns documentation that answers the question, with a source link for each part. No API key is needed. If nothing matches, it says so.
Introduction: The Leap Brought by Language Models and the Limits of Its Impact
No two people would disagree that artificial intelligence models have brought about a qualitative leap in our contemporary lives, or that they will greatly affect the future of humanity. Some jobs will disappear and others will be born. There is also catastrophic environmental harm, which so far has received little attention, though the outcry will grow louder in the coming years. This leap is due primarily to large language models that understand human language. They are creative and inventive, with a formidable capacity to store information and extract what is needed from it. Only a few years ago, this was closer to science fiction than to reality. These models help us a great deal in our daily work across many fields. But to what extent can these models (or tools, as I like to call them) help us? And where are the limits at which they stop?
Knowledge journeys on this subject
Scan a code to open the journey on Basira.
Striving Against the Self
Is jihād al-nafs (striving against the self) a harsh struggle or a journey of awareness? Discover how tazkiyah (purification of the soul) brings you back to God through balanced steps.
Defining the Tools and the Essential Difference from Human Beings
To answer these questions, we must first define these tools objectively and scientifically. These tools are distinguished by linguistic abilities that, throughout history, were unique to human beings and set them apart. We must therefore determine the difference between these tools and human beings, so that we can determine the limits of what the tools can do. This leads us to philosophical questions debated among thinkers and scientists, such as: what is intelligence? What is the human being? Without entering into the complexities of philosophical terminology, the essential difference between human beings and these tools is perception, feeling and choice, and choice is the most important factor. A day may come when I do not wish to speak with anyone, or to answer a particular question. But a day will never come when a language model such as ChatGPT refuses to answer my question, unless it has been programmed to do so.
Controversial Issues and the Limits of Relying on Models
This difference between human beings and language models is essential for subjective topics that are open to debate and not universally agreed upon, foremost among them political, philosophical, historical, religious and psychological topics. Through perception and the capacity for choice, a person determines their position on the information they receive or have received in the past, such as what they were raised on, and so breaks free of its hold. They can scrutinise it, research it and ask about it, so as to grasp it with their very being [CHECK: rendering of «ليدركها بوجوده»], ascertain its truth and determine their position on it. This is impossible for language models. Their answers depend primarily on the data they were trained on and the question put to them. So if I ask one about a historical topic, such as the Crusades against Bilād al-Shām (the Levant), it will give me the narrative it has learned. And if it has been trained on several contradictory narratives, it will give me a different answer each time, or an answer that suits my wishes and my position.
Citing Models in Disputed Questions
In short, the limits of language models lie at any controversial topic on which there is no human consensus. Even if a model gives an answer that supports or reinforces my position, this neither strengthens nor weakens my opinion, and it means nothing whatsoever. Quite simply, these models are a tool unsuited to such uses. Paradoxically, many people cite them to support their opinions on controversial topics that may seem self-evident at first glance, such as the right of a people to resist an occupier. Such citation has no value. First, it comes from a linguistic tool that is biased towards what it has learned and towards the way the question was put to it. Second, these self-evident topics are in fact controversial topics on which we as human beings do not agree, and they relate to our very nature and our desires. They are therefore beyond the capacity of these tools. And if you want my opinion on the matter, for some people this right is relative, depending on who the occupier is and who the resister is. The same applies to many matters in our daily lives, where I claim a right that I deny to others. This relates to the human self and its desires, and to a person's intellectual background and existential outlook.
Using Them as a Therapist: Practical Cautions
Another mistaken use of language models is to treat them as a therapist or psychological counsellor, a friend to whom we confide our secrets and problems. The problem here is that these matters depend primarily on feelings, and a language model is entirely devoid of feelings and emotions, however much its makers try to portray it otherwise or work to develop it in this respect. A language model is the result of arithmetic operations of multiplication and division, no more and no less, and it has no share whatsoever in human feeling. In addition, it is biased towards whatever we say and tell it, and it is excessively positive towards the person speaking to it. Consulting it, therefore, even if it helps in the short term, is harmful in the long term. It is like asking the neighbourhood grocer about the cause of your back pain and asking him for a prescription. Here parents must pay close attention to their children. They must give them security and love, give them trust and talk with them. Otherwise, without anyone noticing, the children will become psychological victims of consulting a language model.
In Scientific Topics: Verification Before Adoption
The discussion so far has introduced the limitations of language models in topics that are open to debate and on which there is no consensus. But this does not mean they can be relied upon absolutely in scientific and objective topics, especially when the topic we are researching is new to us and we have no expertise in it. These models can err for many reasons, including that they may not have learned the information or did not find it on the internet. And they will not say "I don't know"; instead, they will give us wrong answers. These answers are the result of complex computations tied to the data the models were trained on and to the constraints set by their programmers. Therefore, we as users must verify that the answers are correct before adopting and acting on them. In general, it is preferable to acquire at least a minimum of information about the topic we are asking about, so that we can judge whether the answer is accurate or not. It is better not to use these models as a final source of information, especially in sensitive, high-risk topics such as medicine. On the other hand, we should make use of these tools' capabilities to increase our effectiveness and productivity across various applications and fields: preliminary research, writing, summarising, proofreading, translation, brainstorming, low-risk creative tasks, help in explaining and simplifying complex concepts…
Conclusion
In conclusion, language models are the revolution of the age, and we can benefit from them a great deal. Indeed, we should make use of them to improve and develop our work and our society. But we must always bear in mind that these models are tools with specific uses. Using them where they do not belong causes many harms, foremost among them misinformation, distortion, and the manipulation of minds and facts. Here the role of baṣīra (insight) comes to the fore, helping us judge when and how to use these tools. A tool without insight may mislead, but a human being with insight can put the tool in its proper place.
basira.center/en/journeys/inner-struggle
The Beauty of Evil
Does evil have a beautiful face? A journey that uncovers the secret of trials and their purposes, inspired by the book *Beyond Evil* (*Mā Warāʾ al-Sharr*) by Shaykh Miṣbāḥ Yazdī. [CHECK: confirm the established English title of the book, if one exists]
basira.center/en/journeys/beauty-of-evil
Nights of Stillness, Days of Giving
Does your sleep affect your soul and your productivity? A journey that uncovers the secrets of night and day, the midday nap (qaylūla), and a sleep programme that will change your life.
Basira
