Urdu ka Janazah
Faizan Usmani decries the lack of language proficiency among Urdu copywriters.
☰
Urdu, Identity and AI
AI has been heralded as a new computing paradigm – a terminology that rivals the big moments in computing history – the introduction of the graphic user interface, the internet and smartphones. What is it that AI presents that makes it a paradigm shift and what catalysed this shift recently?
AI is a broad term used to refer to a range of techniques. Recent advancements often refer to a subset of techniques also grouped as generative AI. Broadly, these constitute the ability of computers to synthesise new content, such as images or language. Some of the techniques in use today date back decades, but consistent advancements have led to breakthroughs. In particular, the last few years have seen the emergence of new mathematical techniques, increases in processing power and available data, as well as the popularity of new interface experiments (the chatbot) that have captured the world’s imagination. We are now envisioning computers that can perform complex linguistic tasks, understand what humans mean more accurately, and replace humans in many tasks we would not have thought possible before.
In simple terms, AI techniques allow computers to identify patterns in data.
Humans can do this too, but computers can conduct mathematical procedures over large amounts of data much faster. This makes computers uniquely able to find patterns in places where humans may not be able to (as well as learn to mimic existing data). By reading large amounts of text, computers can learn how to find patterns between words. By using this ability, we can make predictions and classifications. For example, guess what the next word in a line of text could be, or identify the emotion expressed in a sentence. We can also use this ability to generate new text.
This new wave of AI capabilities has hit a tech world that is largely centred in the West and in China, where technologists have begun to imagine new opportunities in terms of how computers can help humans. However, I worry that our vision of how to use AI is so centred on Western life and culture that we in Pakistan may benefit less from the technology than we would otherwise do.
When my work on building better technology for written Urdu began, I found it glaring how many technologies, which we consider table stakes for English, are missing for Urdu and other regional languages. Take spell check, which corrects erroneously typed language. This ensures linguistic standards and creates a sense of reliability and credibility in the text, and I assumed this technology would be of extreme value to Urdu publishers.
However, in my discussions with professionals in the Urdu language, I found that automated spell check did not hold the allure that I had expected it to.
Copyediting from professionals is readily available, and publishers value the in-person networks and expertise that their existing mechanisms enable. Would spell check be helpful? Sure. But it is not the technology that is holding back the progress of the Urdu language. In fact, in some instances, it may negatively impact the culture of publishing, and this is why it is important to understand the cultures in which our languages and our people operate.
Precedents for the sort of shorthand analysis I conducted are available throughout history. When the printing press was introduced in Europe, Muslim printers did not adopt it straight away. Historians first assumed this was because the Muslim world was backward. However, our assumptions about this have since evolved. It now appears more likely that the printing press was slow to be adopted because the needs of the Muslim world at the time were not to spread the written word across geographies quickly. Muslim education at the time relied on in-person teaching and apprenticeship. What this story illustrates is that while the benefits of technology may be available to all cultures and people, different regions require different things at different times.
Here is another short story that is illustrative of how our cultural blindness can limit our ability to progress in technology. In my role as an educator and mentor to young designers across Pakistan, I have found that the core of our professionals are more familiar with the principles of Western design and typography than they are with typography in the Arabic script or South Asian design principles. This is because we are short of professionals with expertise in our native design disciplines while being at the same time proficient in the vocabulary of Western higher education, which big-name Pakistani educational institutions draw from.
The result is often the belief that design knowledge (and even high culture, aesthetics and taste) is the domain of the West and not of Pakistan.
The truth, however, is that, as far as I can see, Pakistan is the country that cares the most about typography in the world.
Pakistan prints Urdu in the Nastaliq script. The aesthetic of Nastaliq is so closely tied to the Urdu language that until the late 1980s, Pakistani newspapers preferred to handwrite and lithograph rather than use movable type, which was introduced in Europe many centuries ago. Time and again, we see examples of how Western visions of technology do not address our needs in the same way. Instead of believing that this somehow makes us technologically backwards, we must make our technology more culturally aware and forward-thinking.
There are some problems with AI tools that are simple to understand. Because these tools train on high-quality data from the West, they are better at understanding Western culture than ours. For example, image generation tools are much better at mimicking the style of Rembrandt than they are at mimicking Sadequain. While technology is becoming cheaper and more widely available, we cannot argue that this kind of structural inequality will decrease in the future.
There is also a more complex problem – the fact that our best-in-class vision of AI is centred on Western culture. The jobs we look to automate are jobs considered mundane in the West, such as form filling and copy editing. The magical scenarios we are given are from English science-fiction movies. These are not stories about the Pakistani people, and in chasing them blindly, we may forget the needs of our people. Where is the gadget that automatically turns on our old geyser when the gas is back? Where is the cheap sensor that detects unhealthy levels of toxic fumes from cheap fuels in home kitchens? Where is the monitoring system that collates unexplained viruses killing our loved ones?
Where is the software that digitises old Urdu texts to modern formats? And where are the games that teach our kids about running a tandoor?
The truth is that technologists like me can be smart, but we can also be very dumb. We had to develop a whole new discipline called human-centred design to get around our own failures. This new discipline is rooted in the idea that technologists are very bad at predicting what humans will do, and the best way to design new products is to observe our audience, try out ideas and see how people react to them. It is this approach that needs to be at the core of how we use AI to build new products for Pakistan.
Zeerak Ahmed is Principal UX Designer, Amazon and Founder, Matnsaz. He tweets at @zeerakahmed
Read Comments
Related Stories
Humans can do this too, but computers can conduct mathematical procedures over large amounts of data much faster. This makes computers uniquely able to find patterns in places where humans may not be able to (as well as learn to mimic existing data). By reading large amounts of text, computers can learn how to find patterns between words. By using this ability, we can make predictions and classifications. For example, guess what the next word in a line of text could be, or identify the emotion expressed in a sentence. We can also use this ability to generate new text.
This new wave of AI capabilities has hit a tech world that is largely centred in the West and in China, where technologists have begun to imagine new opportunities in terms of how computers can help humans. However, I worry that our vision of how to use AI is so centred on Western life and culture that we in Pakistan may benefit less from the technology than we would otherwise do.
When my work on building better technology for written Urdu began, I found it glaring how many technologies, which we consider table stakes for English, are missing for Urdu and other regional languages. Take spell check, which corrects erroneously typed language. This ensures linguistic standards and creates a sense of reliability and credibility in the text, and I assumed this technology would be of extreme value to Urdu publishers.
However, in my discussions with professionals in the Urdu language, I found that automated spell check did not hold the allure that I had expected it to.
Copyediting from professionals is readily available, and publishers value the in-person networks and expertise that their existing mechanisms enable. Would spell check be helpful? Sure. But it is not the technology that is holding back the progress of the Urdu language. In fact, in some instances, it may negatively impact the culture of publishing, and this is why it is important to understand the cultures in which our languages and our people operate.
Precedents for the sort of shorthand analysis I conducted are available throughout history. When the printing press was introduced in Europe, Muslim printers did not adopt it straight away. Historians first assumed this was because the Muslim world was backward. However, our assumptions about this have since evolved. It now appears more likely that the printing press was slow to be adopted because the needs of the Muslim world at the time were not to spread the written word across geographies quickly. Muslim education at the time relied on in-person teaching and apprenticeship. What this story illustrates is that while the benefits of technology may be available to all cultures and people, different regions require different things at different times.
Here is another short story that is illustrative of how our cultural blindness can limit our ability to progress in technology. In my role as an educator and mentor to young designers across Pakistan, I have found that the core of our professionals are more familiar with the principles of Western design and typography than they are with typography in the Arabic script or South Asian design principles. This is because we are short of professionals with expertise in our native design disciplines while being at the same time proficient in the vocabulary of Western higher education, which big-name Pakistani educational institutions draw from.
The result is often the belief that design knowledge (and even high culture, aesthetics and taste) is the domain of the West and not of Pakistan.
The truth, however, is that, as far as I can see, Pakistan is the country that cares the most about typography in the world.
Pakistan prints Urdu in the Nastaliq script. The aesthetic of Nastaliq is so closely tied to the Urdu language that until the late 1980s, Pakistani newspapers preferred to handwrite and lithograph rather than use movable type, which was introduced in Europe many centuries ago. Time and again, we see examples of how Western visions of technology do not address our needs in the same way. Instead of believing that this somehow makes us technologically backwards, we must make our technology more culturally aware and forward-thinking.
There are some problems with AI tools that are simple to understand. Because these tools train on high-quality data from the West, they are better at understanding Western culture than ours. For example, image generation tools are much better at mimicking the style of Rembrandt than they are at mimicking Sadequain. While technology is becoming cheaper and more widely available, we cannot argue that this kind of structural inequality will decrease in the future.
There is also a more complex problem – the fact that our best-in-class vision of AI is centred on Western culture. The jobs we look to automate are jobs considered mundane in the West, such as form filling and copy editing. The magical scenarios we are given are from English science-fiction movies. These are not stories about the Pakistani people, and in chasing them blindly, we may forget the needs of our people. Where is the gadget that automatically turns on our old geyser when the gas is back? Where is the cheap sensor that detects unhealthy levels of toxic fumes from cheap fuels in home kitchens? Where is the monitoring system that collates unexplained viruses killing our loved ones?
Where is the software that digitises old Urdu texts to modern formats? And where are the games that teach our kids about running a tandoor?
The truth is that technologists like me can be smart, but we can also be very dumb. We had to develop a whole new discipline called human-centred design to get around our own failures. This new discipline is rooted in the idea that technologists are very bad at predicting what humans will do, and the best way to design new products is to observe our audience, try out ideas and see how people react to them. It is this approach that needs to be at the core of how we use AI to build new products for Pakistan.
Zeerak Ahmed is Principal UX Designer, Amazon and Founder, Matnsaz. He tweets at @zeerakahmed
Read Comments
Related Stories
Copyediting from professionals is readily available, and publishers value the in-person networks and expertise that their existing mechanisms enable. Would spell check be helpful? Sure. But it is not the technology that is holding back the progress of the Urdu language. In fact, in some instances, it may negatively impact the culture of publishing, and this is why it is important to understand the cultures in which our languages and our people operate.
Precedents for the sort of shorthand analysis I conducted are available throughout history. When the printing press was introduced in Europe, Muslim printers did not adopt it straight away. Historians first assumed this was because the Muslim world was backward. However, our assumptions about this have since evolved. It now appears more likely that the printing press was slow to be adopted because the needs of the Muslim world at the time were not to spread the written word across geographies quickly. Muslim education at the time relied on in-person teaching and apprenticeship. What this story illustrates is that while the benefits of technology may be available to all cultures and people, different regions require different things at different times.
Here is another short story that is illustrative of how our cultural blindness can limit our ability to progress in technology. In my role as an educator and mentor to young designers across Pakistan, I have found that the core of our professionals are more familiar with the principles of Western design and typography than they are with typography in the Arabic script or South Asian design principles. This is because we are short of professionals with expertise in our native design disciplines while being at the same time proficient in the vocabulary of Western higher education, which big-name Pakistani educational institutions draw from.
The result is often the belief that design knowledge (and even high culture, aesthetics and taste) is the domain of the West and not of Pakistan.
The truth, however, is that, as far as I can see, Pakistan is the country that cares the most about typography in the world.
Pakistan prints Urdu in the Nastaliq script. The aesthetic of Nastaliq is so closely tied to the Urdu language that until the late 1980s, Pakistani newspapers preferred to handwrite and lithograph rather than use movable type, which was introduced in Europe many centuries ago. Time and again, we see examples of how Western visions of technology do not address our needs in the same way. Instead of believing that this somehow makes us technologically backwards, we must make our technology more culturally aware and forward-thinking.
There are some problems with AI tools that are simple to understand. Because these tools train on high-quality data from the West, they are better at understanding Western culture than ours. For example, image generation tools are much better at mimicking the style of Rembrandt than they are at mimicking Sadequain. While technology is becoming cheaper and more widely available, we cannot argue that this kind of structural inequality will decrease in the future.
There is also a more complex problem – the fact that our best-in-class vision of AI is centred on Western culture. The jobs we look to automate are jobs considered mundane in the West, such as form filling and copy editing. The magical scenarios we are given are from English science-fiction movies. These are not stories about the Pakistani people, and in chasing them blindly, we may forget the needs of our people. Where is the gadget that automatically turns on our old geyser when the gas is back? Where is the cheap sensor that detects unhealthy levels of toxic fumes from cheap fuels in home kitchens? Where is the monitoring system that collates unexplained viruses killing our loved ones?
Where is the software that digitises old Urdu texts to modern formats? And where are the games that teach our kids about running a tandoor?
The truth is that technologists like me can be smart, but we can also be very dumb. We had to develop a whole new discipline called human-centred design to get around our own failures. This new discipline is rooted in the idea that technologists are very bad at predicting what humans will do, and the best way to design new products is to observe our audience, try out ideas and see how people react to them. It is this approach that needs to be at the core of how we use AI to build new products for Pakistan.
Zeerak Ahmed is Principal UX Designer, Amazon and Founder, Matnsaz. He tweets at @zeerakahmed
Read Comments
Related Stories
Pakistan prints Urdu in the Nastaliq script. The aesthetic of Nastaliq is so closely tied to the Urdu language that until the late 1980s, Pakistani newspapers preferred to handwrite and lithograph rather than use movable type, which was introduced in Europe many centuries ago. Time and again, we see examples of how Western visions of technology do not address our needs in the same way. Instead of believing that this somehow makes us technologically backwards, we must make our technology more culturally aware and forward-thinking.
There are some problems with AI tools that are simple to understand. Because these tools train on high-quality data from the West, they are better at understanding Western culture than ours. For example, image generation tools are much better at mimicking the style of Rembrandt than they are at mimicking Sadequain. While technology is becoming cheaper and more widely available, we cannot argue that this kind of structural inequality will decrease in the future.
There is also a more complex problem – the fact that our best-in-class vision of AI is centred on Western culture. The jobs we look to automate are jobs considered mundane in the West, such as form filling and copy editing. The magical scenarios we are given are from English science-fiction movies. These are not stories about the Pakistani people, and in chasing them blindly, we may forget the needs of our people. Where is the gadget that automatically turns on our old geyser when the gas is back? Where is the cheap sensor that detects unhealthy levels of toxic fumes from cheap fuels in home kitchens? Where is the monitoring system that collates unexplained viruses killing our loved ones?
Where is the software that digitises old Urdu texts to modern formats? And where are the games that teach our kids about running a tandoor?
The truth is that technologists like me can be smart, but we can also be very dumb. We had to develop a whole new discipline called human-centred design to get around our own failures. This new discipline is rooted in the idea that technologists are very bad at predicting what humans will do, and the best way to design new products is to observe our audience, try out ideas and see how people react to them. It is this approach that needs to be at the core of how we use AI to build new products for Pakistan.
Zeerak Ahmed is Principal UX Designer, Amazon and Founder, Matnsaz. He tweets at @zeerakahmed
Read Comments
Related Stories
The truth is that technologists like me can be smart, but we can also be very dumb. We had to develop a whole new discipline called human-centred design to get around our own failures. This new discipline is rooted in the idea that technologists are very bad at predicting what humans will do, and the best way to design new products is to observe our audience, try out ideas and see how people react to them. It is this approach that needs to be at the core of how we use AI to build new products for Pakistan.
Zeerak Ahmed is Principal UX Designer, Amazon and Founder, Matnsaz. He tweets at @zeerakahmed