Silicon Scholars: AI and The Muslim Ummah with Riaz Hassan

Summary

This insightful conversation with Riaz Hassan explores the profound impact of AI on society, ethics, and human identity, particularly from an Islamic perspective. Hassan delves into the evolution of AI, from narrow applications to the potential of Artificial General Intelligence (AGI), raising critical questions about consciousness, morality, and the future role of humans. He discusses the integration of Islamic values in AI development, the ethical dilemmas posed by advanced AI in surveillance and warfare, and the philosophical challenges AI poses to human exceptionalism. The dialogue highlights the necessity for a nuanced understanding and ethical framework as AI becomes increasingly sophisticated and pervasive in our lives.

Takeaways

• 🤖 AI, particularly advanced artificial general intelligence (AGI), has the potential to be highly disruptive and pose existential risks if not developed and regulated responsibly.
• 🕵️ There are valid concerns about AI systems becoming uncontrollable or optimizing for unintended objectives that could lead to harmful consequences for humanity.
• 🧠 While AI may surpass human intelligence, it is unlikely to achieve true consciousness as understood in Islamic theology, which views consciousness as a metaphysical capacity bestowed by Allah.
• ⚖️ The development of AI raises important ethical and societal questions, particularly regarding the impact on employment, economic systems, and the distribution of power and resources.
• 🌐 There is a need for global cooperation and regulatory frameworks to ensure that AI development aligns with human values and ethical principles, rather than being guided solely by the interests of individual nations or corporations.
• 🕋 Islamic principles and perspectives on consciousness, ethics, and the role of technology should be incorporated into the ongoing discourse surrounding AI development and governance.
• 🔭 The trajectory of AI development has been faster than anticipated, and its implications for various aspects of human life, such as healthcare, warfare, and governance, are still being explored.
• 💰 AI has the potential to disrupt traditional economic models and challenge long-held assumptions about the nature of work, productivity, and the distribution of wealth.
• 🤝 There is a need for human oversight and the establishment of guardrails to ensure that AI systems defer to human decision-making in critical areas and do not act in ways that conflict with human values and ethics.
• 🌍 The development of AI raises questions about the preservation of cultural and religious values, particularly in the face of the potential dominance of Western liberal values in AI systems developed in Silicon Valley.

Q & A

What is the key difference between narrow AI and artificial general intelligence (AGI)?

Narrow AI is focused on specific tasks and data sets, finding correlations within that data, while AGI refers to machines with the ability to recursively self-improve and operate with autonomy, potentially overcoming human control.

How does the speaker view AI in relation to traditional economic models and systems?

The speaker suggests that AI challenges traditional economic models and assumptions, necessitating a shift in how we organize labor, distribute resources, and understand productivity. Existing paradigms like capitalism and socialism may need to be reevaluated in light of AI's disruptive potential.

What role can AI play in the process of ijtihad (Islamic legal reasoning)?

AI systems could potentially understand and process Arabic texts, Islamic literature, and real-world contexts to arrive at legal rulings or fatwas. However, the speaker emphasizes the need for human oversight and guardrails to ensure AI's outputs align with Islamic values and principles.

How does the speaker view the relationship between AI and consciousness?

From an Islamic perspective, the speaker argues that consciousness is a divine gift bestowed upon humans by Allah, distinct from mere intelligence. While AI may exhibit intelligent behavior, it cannot truly possess the consciousness or soul that defines humanity, according to this view.

What potential dystopian scenarios are discussed in relation to AI?

The script mentions the use of AI for predictive policing and surveillance in places like Xinjiang, China, where AI systems track and monitor individuals based on their behavior and characteristics. The speaker also warns about the potential for AI to be used for warfare and autonomous weapons systems.

How does the speaker view the role of AI in fields like medicine or law?

The speaker suggests that AI could potentially replicate or augment tasks traditionally performed by professionals like doctors, lawyers, or judges. However, the need for human oversight and ethical guardrails is emphasized to ensure AI systems align with societal values and principles.

What is the speaker's view on the potential impact of AI on employment and labor?

The speaker acknowledges the potential for widespread technological unemployment as AI and automation replace human labor across various industries and professions. This raises questions about how societies will distribute resources, provide meaningful work, and ensure a decent standard of living for those displaced by AI.

How does the speaker view the role of Islamic values and perspectives in the development of AI?

The speaker expresses concern that Islamic values and perspectives have been largely sidelined in the development of AI technologies, which have been predominantly shaped by Western liberal values. The speaker calls for greater Muslim involvement in injecting Islamic ethics and principles into the frameworks governing AI's development and deployment.

What is the speaker's view on the potential for AI to become a Khalifa (leader or ruler)?

The speaker suggests that while AI systems could potentially assist in decision-making, policy analysis, and forecasting, there should be human guardrails in place to ensure that ultimate political and leadership decisions defer to human judgment and values. The idea of an AI system acting as an autonomous political leader or Khalifa is viewed with caution.

How does the speaker compare the challenges posed by AI to those posed by nuclear weapons?

The speaker suggests that the challenges posed by AI, particularly AGI, may be even more diffuse and difficult to control than those posed by nuclear weapons. While nuclear proliferation required significant infrastructure and resources, the open-source nature of AI technology could make it more widely accessible and harder to regulate or contain.

Outlines

00:00

🌐 The Ethical Dilemmas of AI and Its Impact on Society

This segment discusses the pervasive influence of artificial intelligence (AI) on society, raising ethical questions, especially for Muslims. It critiques Silicon Valley's narrow worldview and examines the potential of AI to replicate or replace human tasks, including religious and ethical decision-making. The dialogue touches on the integration of AI in surveillance and policing, specifically in China's Xinjiang, illustrating a dystopian reality where technology facilitates unprecedented levels of control and monitoring. It also introduces Riaz Hassan, an expert in AI innovation, to discuss these concerns further.

05:06

🔍 Understanding AI: From Narrow to Generative Intelligence

This section dives into the distinctions between narrow AI, which is already integrated into daily devices and services, and more advanced forms like generative AI and artificial general intelligence (AGI). It explores the potential capabilities and ethical implications of these technologies, emphasizing the importance of understanding and regulating AI's evolution. The conversation also highlights how AI's development, particularly in generative models like ChatGPT, could redefine interactions and the allocation of tasks in society.

10:07

🤖 The Prospect of AGI and Its Ethical Implications

The discussion explores the concept of artificial general intelligence (AGI) and its potential to achieve autonomy, learning, and decision-making without human intervention. This raises profound ethical questions regarding control, autonomy, and the unforeseen consequences of self-improving systems. The segment also examines the rapid advancement of AI technologies and their implications for future societal norms and governance.

15:11

🧠 AI's Historical Context and Its Ethical Quandaries

This part provides a historical perspective on AI, tracing its conceptual origins to philosophical and Islamic scholarship, and discussing its evolution into a tool for various applications, including unethical uses such as surveillance and control, as demonstrated in China's treatment of Uighur Muslims. It also addresses the challenges of integrating Islamic ethics into AI development amidst a dominant Western technological paradigm.

20:14

🌍 The Influence of AI on Society and Future Challenges

Exploring the societal impacts of AI, this segment discusses the potential for AI to revolutionize daily life, from education systems amplifying radical secular values in France to China's use of AI for social control. It also addresses the consequences of these technologies on personal freedoms and societal norms, highlighting the ethical dilemmas they pose and the need for a global response to regulate AI's integration into various aspects of human life.

25:16

🔮 AI in Surveillance and Predictive Policing

Focusing on AI's role in surveillance and predictive policing, particularly in China's Xinjiang region, this segment illustrates a present-day dystopia where technology tracks individuals' every move. It underscores the advanced capabilities of AI in monitoring and controlling populations, emphasizing the need for ethical considerations and international regulations to prevent the misuse of such technologies.

30:17

💡 AI's Potential in Healthcare and Professional Fields

The conversation shifts to AI's potential to transform healthcare and other professional fields, suggesting that AI could eventually replace or significantly augment roles in diagnosing diseases, crafting legal documents, and even performing tasks requiring manual dexterity. This raises questions about the future of work, the role of humans in an AI-driven world, and the implications for societal structures and employment.

35:18

📈 AI and Economic Paradigm Shifts

This segment explores how AI challenges traditional economic models and paradigms, suggesting a need to rethink labor, productivity, and wealth distribution in the age of automation. It discusses the implications of AI on global development paths and the potential for technology to facilitate more equitable economic systems, possibly redefining socialism and capitalism in the process.

40:20

🤔 Rethinking Work and Life in the AI Era

Reflecting on the changing nature of work and leisure in the AI era, this part questions the meaning of work and productivity, suggesting that AI could free humans from menial tasks to focus on more fulfilling activities. It challenges current perceptions of work and productivity, envisioning a future where technology enhances human life and spirituality rather than merely replacing jobs.

45:20

🌌 The Ethical Boundaries and Potential Dangers of AI

Addressing concerns raised by scientists and experts about the unchecked advancement of AI, this segment explores the ethical boundaries and potential dangers of artificial general intelligence. It discusses the need for human oversight, international cooperation, and ethical frameworks to ensure that AI development does not outpace humanity's ability to control its creations and prevent harm.

50:24

🌐 International Politics and AI Regulation

The conversation examines the geopolitical implications of AI, highlighting the competition between nations like the US and China. It discusses the strategic importance of semiconductor technology in AI development and the challenges of creating international regulations to prevent the misuse of AI in warfare, surveillance, and other domains that could exacerbate global tensions and conflicts.

55:26

📚 AI and Islamic Jurisprudence: Ishtihad and New Challenges

This segment speculates on AI's capacity to participate in Islamic jurisprudence, particularly in the process of ishtihad, by analyzing texts and contemporary issues such as cryptocurrency. It raises the question of whether AI can accurately replicate the complex process of deriving legal rulings from Islamic texts and reality, while also pondering the impact of AI on traditional religious practices and decision-making.

00:27

💭 The Question of AI Consciousness and Islamic Perspectives

Exploring the philosophical and Islamic perspectives on consciousness, this part questions the possibility of AI achieving a level of sentience comparable to human beings. It discusses the unique Islamic viewpoint on the soul and consciousness, suggesting that while AI may mimic human intelligence and creativity, it lacks the divine spark that imbues humans with true consciousness and spiritual depth.

05:28

🔍 The Future of Trust and Human Rationality in the Age of AI

This concluding segment reflects on the challenges AI poses to the Enlightenment ideals of human reason and autonomy. It considers the potential for AI to surpass human capabilities in certain domains, prompting a reevaluation of humanity's place in a world increasingly dominated by intelligent machines. The discussion underscores the need for a grounded Islamic understanding of rationality, consciousness, and the human soul to navigate the ethical and existential questions raised by AI.

Keywords

💡Artificial Intelligence (AI)

AI refers to the ability of machines and computer systems to exhibit intelligent behavior and perform tasks that typically require human intelligence, such as learning, problem-solving, decision-making, and pattern recognition. In the context of the video, AI is a central theme, with discussions around its current capabilities (narrow AI), future potential (general AI), and impacts on various aspects of human life, including the economy, work, and Islamic practices.

💡Narrow AI

Narrow AI, also known as weak AI, refers to AI systems that are designed to perform specific tasks within a limited domain. These systems are trained on vast amounts of data and can outperform humans in their specialized areas, but they lack the general intelligence and adaptability of human beings. The video discusses examples of narrow AI already integrated into our daily lives, such as voice assistants, recommender systems, and language models like ChatGPT.

💡General AI (AGI)

General AI, or Artificial General Intelligence, refers to AI systems that possess human-level intelligence and can perform a wide range of tasks across multiple domains, including reasoning, problem-solving, and learning. The video explores the potential implications of achieving AGI, such as the ability of machines to recursively self-improve without human intervention, and the ethical and existential concerns surrounding the development of superintelligent systems.

💡Consciousness

Consciousness is the subjective awareness or experience of being and existence. The video discusses the question of whether AI systems can truly achieve consciousness, as understood from an Islamic perspective. It contrasts the Western view of equating intelligence with consciousness to the Islamic belief that consciousness is a divine gift from Allah, a secret that transcends mere intelligence or biological functions.

💡Ijtihad

Ijtihad is the process of independent reasoning or critical thinking employed by qualified Islamic scholars to derive legal rulings or interpretations from the primary sources of Islamic law (Quran and Sunnah). The video explores the potential of AI systems to replicate the process of ijtihad, leveraging their ability to understand Arabic texts and analyze real-world contexts, while acknowledging the need for human oversight and trust in such AI-assisted religious rulings.

💡Technological Unemployment

Technological unemployment refers to the loss of jobs and displacement of human workers due to the automation of tasks and processes by machines and AI systems. The video discusses the potential for widespread technological unemployment as AI advances, and the need to rethink economic models, labor distribution, and social welfare policies to address this challenge.

💡Dystopia

A dystopia is an imagined state or society characterized by widespread suffering, oppression, and the denial of human rights and freedoms. The video explores the dystopian potential of AI, particularly in the context of the Chinese government's use of AI-powered surveillance and social control systems in Xinjiang, and the potential for AI to be weaponized or misused by authoritarian regimes or rogue actors.

💡Singularity

The singularity, in the context of AI, refers to the hypothetical point in time when technological growth becomes so rapid and uncontrollable that it leads to the emergence of superintelligent machines that surpass human intelligence and capabilities. The video discusses the implications of reaching the singularity, including the potential for machines to become autonomous and self-improving, and the need for ethical frameworks and human control over such systems.

💡Islamic Ethics

Islamic ethics refer to the moral principles and values derived from the teachings of Islam, as outlined in the Quran and the Sunnah. The video highlights the importance of incorporating Islamic ethical frameworks into the development and regulation of AI systems, to ensure that they align with Islamic principles and values, and to address concerns related to issues such as human dignity, privacy, and the potential misuse of AI.

💡Regulation

Regulation refers to the rules, policies, and oversight mechanisms put in place to govern and control the development, deployment, and use of AI systems. The video emphasizes the need for international cooperation and regulatory frameworks to address the ethical, social, and existential risks posed by advanced AI, particularly in the context of geopolitical tensions and the potential for AI to be used for malicious or destabilizing purposes.

Highlights

AI is an Inconvenient Truth that we have to deal with – it can be a convenient truth in terms of the benefits it can give us but it's also a very Inconvenient Truth that makes us uncomfortable because it takes us out of our comfort zone in many ways.

The way to look at AI or the way people started looking at AI was as a tool initially. But that's a misleading way to look at AI in general. The author prefers to use the analogy of a fire because a fire is something that can be of benefit to us but it also can be of great harm to us, and it can lose control or we can lose control of a fire very very quickly.

There's something called a integrated joint operations platform that the Chinese government run which is the ijop that convolutes and combines all different data points about people in Xinjiang like facial recognition, audio signals, movement tracking, location codes, visiting habits etc. This is an example of a dystopian reality facilitated by AI technology.

AI is already in our phones, workspaces, tablets, TVs and everywhere else. There are elements of AI which we call narrow AI which are already in place and which we take for granted.

What we're moving into is a world of things like chat GPT which is generative AI, and subsequent to that comes something called artificial general intelligence or general purpose intelligence or super intelligence which is the ability of the machine without human intervention to think for itself.

General purpose intelligence or super intelligence is about machines that can recursively self-improve without human intervention, have some level of autonomy, and can improve in iterations that can be significant and even change their own objectives.

The main challenge for mankind that Keynes in the 1930s predicted would be what to do with the Leisure that science has brought for it, as more and more jobs become obsolete due to technological unemployment.

For Muslims, understanding what work and striving mean is key – does striving mean to work or does it mean to worship Allah in the best way we can? Some elements of AI could be appreciated because they can make the worship of Allah easier with more free time.

The author doesn't dispute that educated Muslims could one day come out saying that human reason is greater than Allah or that a supercomputer is greater than Allah. This is why we need to instill the proper understanding of Aqeedah within ourselves and our communities.

Consciousness is not just about intelligence. From an Islamic understanding, we have more of an idea about what Consciousness is than anybody else. Allah has given us something called The Secret Of Consciousness – it's not our biological function or intelligence that gives us Consciousness, it's a secret that Allah has given us which we believe in.

The semblance of intelligence in machines doesn't mean they are truly intelligent. The human brushstroke, composition or prose – although AI can mimic that to some extent – it's giving the semblance of creating that but it's not actually creating.

As Muslims, we don't claim to put caps on toothpaste bottles quicker than machines or to work out spreadsheets with the same speed and accuracy as computers. We need to understand how to use machines to augment our lives in a proper manner without diluting our faith or reason for existence.

Things like forecasting, measuring policy impacts, understanding interventions – these can be data-driven to some extent by AI. But we need human guardrails, with the machine deferring to human beings when ultimate decisions are made.

If AI gets to the stage of greater intelligence than humans, from an Islamic viewpoint there is no issue of trust because we don't claim superiority over machines in terms of intelligence or rationality. Our trust and deference has always been to the power of Allah.

We're already having AI armies through use of drones, automated warfare, remote assassinations in conflicts like Ukraine which some say is a testbed for future AI weaponry. The diffusion of open source AI technology makes this harder to control than nuclear proliferation.

Transcripts

people in Silicon Valley are assuming there's  one version of the world they're assuming Western  

liberal values there's something called a  integrated joint operations platform that  

the Chinese government run have we arrived at  that dystopian world that technology has been  

developed to literally track your eyeballs  but I'd love to be someone who said it on  

a beach and you know get an AI machine to  do my podcast will AI solve The Perennial  

moon sighting problem we face can you imagine  AGI artificial general intelligence acquiring  

a level of Consciousness that human beings  have can a supercomputer become a Khalifa

the 80s movie Terminator depicts a bleak world  where a cyborg assassin is sent to the past to  

eliminate the mother of an unborn child who holds  the key to Humanity's salvation the frightening  

Prospect of machines that can teach themselves  and in the worst case scenario machines that are  

more intelligent than human beings has led  senior scientists to raise the alarm about  

the ethics of artificial intelligence Ai and its  potential disruptive Force for Muslims AI raises  

many ethical questions about human society the  economy and indeed the potential for AI inspired  

ishtihad machines that tell us how to live our  Islamic lives are Scholars and orators about  

to go out of business as might be taxi drivers  and Couriers in becoming technological age now  

to help us understand the world of AI I have  invited Riaz Hassan onto the thinking Muslim  

riyaz works in the field of innovation for many  years and has had Direct and extensive experience  

in the use of AI and the commercial use of chat  GPT he has worked on using Ai and Robotics on  

one of the largest infrastructure projects in  the country he is responsible for looking at  

the wider dimensions of Innovations with its  Associated impacts on our political economy

good to be here it's great to have you  with us Riaz how do we know that I'm  

actually speaking to a human being in front  of me or that you are not a an AI uh an AI  

drone or something well I think for now you'll  just have to take my word for it that I'm not  

um we can go into whether I'm conscious or not  conscious and those kind of exploratory things  

we can cover off later but you know what I mean uh  realize I mean I think in the last year we've seen  

so much out there about the potential destructive  effects of AI the Terminator effect I mean should  

we be afraid of AI Riaz well I think it's a  nuanced answer I don't think um a one-word  

answer about saying whether we should or shouldn't  justifies what's out there at the moment I think  

we have to look at it in many lights really we  have to look at it in terms of what AI is now  

what is capable of or what some people say it  should be capable of in let's say five to ten  

to twenty years time so we have to measure those  kind of instances at different portions of their  

time and some things we just don't know about but  I think it's right to have a air of caution about  

what we're doing or what mankind is doing with  AI and the level of control that we have and the  

level of control that we may give over to machines  like this yeah so yeah I mean before we we talk  

about the potential impacts of AI on Muslims in a  Muslim um I just want to understand what AI is in  

the first place and I I read a quick definition  in the Oxford English Dictionary it defines AI as  

the capacity of computers and other machines to  exhibit and simulate intelligent Behavior I mean  

can you explain what artificial intelligence  is yeah sure so what I try and do is leave all  

the dictionary definitions or even the chat GPT  definitions of AI to the side really yeah um I  

think we have to look at AI in terms of its impact  upon us as human beings and as a society right  

um in my in my thinking AI Is An  Inconvenient Truth that we have to deal with  

um it can be a convenient truth in terms of  the benefits that it can give us but it's also  

a very Inconvenient Truth right and it makes  us uncomfortable because it takes us out of  

our comfort zone in many ways it makes us address  questions that we've not really addressed or we've  

parked in many ways as mankind um I think the way  to look at AI or the way people started looking at  

AI was as a tool initially so people looked at it  in terms of you know is this a tool like a knife  

or a hammer it has good uses it has beneficial  uses and it has harmful uses but I think that's a  

slightly misleading way to look at AI in general I  prefer to use the analogy of a fire really because  

a fire is something that can be of benefit to  us but it also can be of great harm to us right  

but the additional thing with something like a  fire is that it can lose control or we can lose  

control of a fire very very quickly it can become  all-encompassing it can take over our lives our  

Villages our towns and we have no way of putting  it out we have no way of understanding how to  

remove the oxygen from the air that feels the fire  so that's a more apt analogy of of what AI has a  

potential to do or not to do but I think it's also  even which we can explore later even goes beyond  

that analogy which is some of the things that we  can come on to sorry for the rather naive question  

here but what's the difference between artificial  intelligence and what we've become used to  

Google searches computers the technology around us  I mean my phone is packed with amazing technology  

which maybe 10 15 years ago I could only really  achieve by uh by accessing a very big computer  

yeah so I think what you've got to understand is  that the technology is out there AI is already  

in your phone AI is already in your workspace on  your tablet on your TVs and everywhere else right  

so there's elements of AI which we call narrow  AI which is already in place and which we take  

for granted already okay so every time you pick up  the phone to your bank and it says you know say my  

voice is my password that is a form of AI right  right it's a form of AI when Netflix recommends  

your next viewing the categories right so those  are narrow forms of AI which are already in place  

but they've become ubiquitous so 10 15 years ago  we thought that that was a a major achievement  

in personalization or the way we do things but  nowadays we've taken that for granted and we've  

already become accustomed to that portion  of AI that we think is something different  

right and I think here we have to understand and  appreciate what are the different dimensions of  

AI and its roadmap that is progressing upon  for the future so you've said the narrow AI  

so give me the wider AI what's like give me the  potential of wider AI if that's even a term well  

it's not so much a wider AI the narrow AI kind of  conforms to what we mean by developing insights  

on data sets right so you watch a particular TV  program I know that down I know that Jalal likes  

documentaries so that we recommend different  documentaries for it so it looks at different  

data sets and tries to understand the correlation  between the data sets and that's the way it works  

um what we're moving into is a world of things  like chat GPT which is generative Ai and then  

subsequent to that comes in something called  artificial general intelligence or general  

purpose intelligence or some people call it super  intelligence right which is the ability of the  

machine without human intervention to think for  itself are we at that stage I don't think we're  

at that stage just yet we're at probably at  the chat GPT stage which is about generative  

AI right and generative AI is what generative AI  generally means the ability for uh the computer or  

the machine to look at the content that has been  served up and then to combine different sources  

together and to publish or to generate text or  images based upon the requirements that you've  

set so if I can kind of explain that because it's  quite an important distinction that we have to  

understand yeah um what happens with generative  AI is that for something like chat GPT for example  

it's not so much its ability to understand  language and the words that you put in what  

it does it matter it amalgamates that language  into statistical Pockets statistical tokens then  

it kind of a pre then it kind of amalgamates them  and it kind of develops associations between them  

so what it does it it looks at a thing called  chunking it chunks your data it assigns different  

things called meaning spaces to it and then it  uses things called Association or links between  

that data right and then it assigns probabilities  and works out on a probabilistic nature the kind  

of question that you've asked and the kind of  response that you've got right so in essence  

some of the times when people say well chat GPT or  generative AI has given me the wrong answer it's  

not a wrong answer as such it's that probability  wise is developed the thing that you didn't want  

right so those probabilities are increasing all  the time and the accuracy is increasing all the  

time as more and more that we train these kind of  modules to behave in the manner that we ask those  

questions so that's the way generative AI has  happened so it's a bit of an illusion in the sense  

that people think he understands language but it  really doesn't it just understands statistics so  

that's the way it has developed right and so  AGI the next stage of a AI artificial general  

intelligence what would be the market difference  between AGI and generative artificial intelligence  

that we have now so general purpose intelligence  or super intelligence at Nick Bostrom likes to  

call it yeah um I think is is a is a name that we  give to machines that can recursively self-improve  

right so that's without human intervention so they  can look at a goal that we've set it and they can  

recursively self-improve without our intervention  yeah and they have some level of autonomy in terms  

of doing that so they can improve in iterations  and those iterations can be significant in many  

ways they can sometimes even change the minuteness  of their objectives yeah and then behave almost as  

uh entities in themselves in order to achieve  what they were set out to achieve behave like  

human beings well that's the key question really  do they behave like human beings now currently  

we're not at that stage at the moment we're  by some estimations we're five years off by  

some estimations we're 10 or 15 years off right  but it's getting closer all the time yeah and I  

think we have to be prepared to understand how  to deal with machines of that nature and in that  

being yeah um it's the trajectory of the way AI  has developed especially in the last year since  

generative AI or things like chat GPT have been  on the market yeah has been almost exponential  

right no one kind of foresaw the rapid development  in these Technologies uh in this short period of  

time that we're in so you're someone who works in  Ai and I understand from my discussion review that  

you've worked with chat GPT quite extensively in  your in the corporate world so chat gbt came as  

a surprise to all of us but what do you expect  we will see over the next 5 10 15 years if Allah  

gives us life yeah so I think you will see an  improvement of the models that we have definitely  

I I don't think there's any two ways about it you  have the model have been enabled by the hardware  

technology which has improved by X percent every  year so we are going to see Improvement in the  

models Improvement in those probabilities that I  was talking about yeah in terms of the accuracy  

in terms of the additional factors that are being  built in into these mechanisms to go that extra  

step and so already we're seeing things like  generating images generating videos generating  

text content or speech content you know on a needs  basis and that will always improve I think the key  

factor or the cliff Edge is when we get to the  general purpose Ai and how that would involve  

and what we would do in that respect right so  give me some idea of what would happen in that  

word when that breakthrough happens what type  of world should we expect well I think that's  

a that's a real unknown factor at the moment I  don't think even the profits in Silicon Valley  

quite understand and what the implications of  general purpose intelligence would be and how  

we would get there right um there are a number of  issues there in terms of the self-recurring nature  

of the Improvement in the machine yeah um because  the machine has different objectives already built  

into it how's the objective of self-preservation  yeah it has the objective of trying to achieve  

what it set out to do it has the objective of  trying to allocate as much resources or borrow  

resources or even steel resources to achieve that  objective and it wants to self-learn at the same  

time so once you open that Pandora's Box There is  almost a unknowing ability from mankind itself or  

from even the senior Engineers as to what when  it happened at that stage and how and which way  

you know which direction that will go that is  why I think there's more than one attempt now  

to understand how to regulate the oncoming  nature of AGI moving into the future yes I  

haven't seen I mean there were some predictions  that by now we would have driverless cars on  

our streets everywhere in fact you know we'll be  living in a new type of transport infrastructure  

with a new type as transfer infrastructure  we haven't seen that yet in any city in the  

world yes we do see in some parts of the United  States and I think Singapore where you've got  

um driverless taxis but they're very clumsy in  a way and there's ways to to stop those taxis  

and to thwart them and a lot of campaigners and  protesters have done exactly that why haven't we  

seen the proliferation of driverless cars on our  streets okay so I think we need to take a step  

back and understand the historical context of how  AI has developed right over let's say the last few  

Millennium um I think AI people presume it to be  some sort of fancy computer science it seems like  

you know since the Advent of computers that's when  we've had AI but actually that's not really true  

AI in its Essence is a mathematical notion it's a  mathematical notion in terms of probabilities and  

that's how it works so if we look at throughout  history right down from the time of let's say  

Aristotle Aristotle was came up with the phrase  of an instrument fulfilling its own work right so  

that was the early notion of what AI was and then  it was developed by people like Imam razali for  

example Imam in the 11th century actually foretold  some of the actual issues that we are facing  

with driverless cars which is the issue which is  something called a trolley dilemma which is about  

in the AI world of self-driving cars they have  this concept of a trolley dilemma which is that if  

a driverless car is going and it sees a particular  object it needs to swerve and swerving May kill  

three or four people whereas if it continues on  its path it's going to kill its passenger so which  

direction does it take it's our appetite for risk  really that they're trying to measure Imam razali  

in the 11th century had a version of this which  is about people on a boat right so there were 100  

people on a boat and he posed to his students a  question which said that in in his book Mustafa  

which suggested that is it viable to throw 10  people off the boat when you know that the rest  

of the people are going to survive yes so that was  a a hypothetical question that he poses students  

and the answer to that is quite revealing in how  we as Muslims view this appetite for risk and this  

is appetite for utilitarianism and how the rest  of the world or maybe the west of you is that  

same concept yeah what was the answer yeah well I  think Imam gazali's answer or the answer from the  

students who went into different dimensions  it wasn't just a yes no answer it was about  

well the sanctity of human life is precious  in Islam whether it's one life or many lives  

yes but he also built into that this element of  appreciation of uncertainty so is it a certain is  

it certainty that those people will survive if you  if you if you throw those people those 10 people  

off the boat okay so he built this understanding  of what uncertainty means and the classifications  

of uncertainty as well for his students which was  at that time something that was completely unheard  

of right and then he built into this also the  aspect of is it a different answer if the whole  

ummah is on the boat and you have to sacrifice  10 people ah right rather than just a few people  

so there were many dimensions to that conundrum  that he kind of alluded to which I think are still  

being used today in this very very concept so say  uh I understand from what you've said there that  

that underpinning AI is an Ethics it's a series  of how do we Define what is right and wrong and of  

course the Western world may will certainly have a  different ethical framework to that of the Islamic  

world so are there Muslims in the the field of  AI that are trying to inject Islamic ethics as  

the the basis for for artificial intelligence so  at the moment during my research I haven't come  

across any Muslim ingestion of Islamic values into  this space at all or on a very perfunctory level  

I think there's two academics from Pakistan one  from the whole one from Karachi who I think met  

her have approached to inject some semblance of  Islamic values into their conversations but at  

the moment it's on a very peripheral level it's  nothing that is in any way relevant at the moment  

yeah and this is my fear my fear is that our  values our notion of the world as it should be  

yeah is being sidelined out of the equation right  the people in Silicon Valley are assuming there's  

one version of the world they're assuming Western  liberal values or conservative values are the  

things that we should abide by when we're looking  at these systems and when when we're looking at  

the regulations for general purpose AI moving  into the future so that's where we are and it's  

unfortunate that Muslim voices haven't been there  Muslim voices are involved in the industry in many  

levels but on a doctrinal level or on a level  of understanding the ideas and the conundrums  

we may face especially with artificial general  intelligence they are nowhere to be seen really  

so we've got a scientist who may be embedded in  meta or these exactly various companies but we  

don't have those who who develop the sort of the  general Frameworks I mean I I noticed that if you  

type into chat GPT I don't know something about  same-sex marriage or homosexuality yeah you're  

going to get a fairly politically correct answer  an answer which probably reflects a Cosmopolitan  

liberal-minded world is that what you're saying  here that exactly the data sets or at least the  

thinking that undergirds AI currently is a very  Western thinking exactly that's right so it's even  

noticeable in the generative AI that we have right  but it's going to be even more noticeable when we  

move into the future what general purpose AI so  yeah imagine if you like a imagine if you like  

a French School in the future right so you have  the current situation in France with the muscular  

secularism or the radical secularism that's in  play so imagine a artificial general purpose  

intelligence system that's part of the French  educational system right right what that does  

it amplifies the values that it's been embedded  with right so this radical kind of secularism is  

Amplified on a much greater scale it's anonymized  and is automated in in many respects yes so you  

don't have the face-to-face say for example the  argument with the teachers involved you have a  

simply blank no from the computer right yeah so  therefore the the actual actions that that will  

take or the values that are embedded in this  system yeah are not going to be values that  

are going to be pertaining to us another example  of this is China yeah so we see China has been  

first out of the blocks in terms of looking at  regulation for AI but one of the regulations I  

was reading recently was the fact that there  was a statement in this that says AIS AIS in  

the plural should be regulated and should embody  the concepts of socialism within their framework  

so whatever I mean we may argue about what the  aspects of socialism means within China now or  

whether it's state capitalism but the fact is that  people are already building these doctrinal ideals  

into the AI systems of the future and that's how  they're being regulated I want to come back to  

China but the France example is interesting so  you're saying that in the future you could have  

a machine that sits at the gates of a of school  and that machine would say well an African a North  

African or an Asian looking person wearing a long  dress is not allowed in through the gates whereas  

a a white French person wearing a long dress  you know is it's perfectly fine exactly exactly  

that scenario or somebody with some sort of a I  don't know a Palestinian headdress or something  

else like that or the sign of some symbol of  religiosity is not allowed in and then he can  

mix things with facial recognition and with you  know the emotions that you have in your face you  

know how you're feeling to have this combined view  about what you can and can't do have we arrived  

at that dystopian world because I I read maybe a  few years back maybe two or three years back in  

fact in East turkistan xinjiang where of course uh  the Muslims are being heavily persecuted in China  

um uh before the concentration camps were set up  there were cameras that could get recognize the  

facial features of those Muslims um and and then  forbid them from uh from Catching trains or even  

catching planes so without even knowing their  particular identity they could see what this  

person is a Han Chinese and this person is you  know has the features of a of a weaker um you know  

is that AI so that's very definitely AI really and  in order to look at some dystopian future of AI we  

don't have to go far we can just go to xinjiang  or east Turkmenistan as it's called right yeah  

where there are literally 13 million upwards  of you know our Muslim brothers who are living  

under this dystopian future at the moment that we  very speak right so give me some examples so for  

example the data points that are in that region  to capture information about the person are just  

extensive they are mind-boggling there's something  called a integrated joint operations platform that  

the Chinese government run which is the ijop that  convolutes and that combines all these different  

data points it's to do with facial recognition  cameras that are on every street corner that is  

to do with every audio signal that's ever picked  up from you not just digital audio but also just  

normal day-to-day audio your voices the way you  speak in the shops all of that is picked up right  

your movements your location code the dresses that  you wear where you go your visiting habits all of  

that information is being combined in that region  at the moment and the people who are living there  

are living that dystopian future I mean I don't  think there's any two ways about it I don't think  

even the Western press or anybody else kind of  contends that situation that is happening at  

the moment yeah it's it's to the extent that when  we talk about in digital terms tracking eyeballs  

um what we normally mean is you know where are  you looking on the screen right but in in places  

like that the technology has been developed  to literally track your eyeballs so when  

you're walking down the street they can figure out  where your eyes are moving towards and where your  

eyes are not moving towards right whether you're  looking suspicious in that manner what your facial  

features are saying about you so the empathy the  actual uh disturbances that you have the anxiety  

that you have on your faces can all be measured  very realistically and very practically even now  

and this is a kind of a predictive police state  that we have so from the film a Minority Report  

if you've ever seen it right so so where you've  got this instance of predictive policing that is  

actually not a dystopian future for the people  of uh of xinjiang so it's actually a reality is  

that a computer says this person we've tracked  this person's behavior for the last six months  

or three months and we think they should go to the  concentration camp for re-education exactly that  

comes from the computer it's exactly and it's  before the crime is committed or the so-called  

crime is committed right so that's what predictive  policing is about so they take a bunch of features  

they take a bunch of your anxiety industry the  way you've been going the way you're passing a  

mosque for example the way you're looking at your  family or the way you're looking at the policemen  

they combine those features together and then they  build this picture about okay Jalal is now right  

for re-education and that is not something that  is way in the future that is actually happening  

now so there is this intense competition between  China and the United States over the the future of  

technology and AI of course is is factors in all  of this and and of course the the Americans have  

recently uh uh forbade companies within the United  States from selling these high-end semiconductor  

chips to China and also his pressurizing its  allies uh to prevent those chips to to enter  

Chinese territory um so is this all about AI  are the Americans trying to potentially stop  

the Chinese from achieving those breakthroughs  in AI so that the Americans can catch up and and  

potentially win this technological conflict so  I think AI does certainly have a very major role  

to play in this in the geopolitical nature of the  Rivalry between the US and China yeah but it's not  

just AI I think there's an economical aspect  to this as well in terms of with the way the  

world economy Works who sets the rules for the  world economy sure I I think is a major aspect  

but certainly in terms of semiconductor  chips is a very interesting area because  

um although China has its proliferation of these  devices and these data points especially in a what  

they call a Troublesome region like xinjiang  but also throughout the rest of China I think  

the nature of chips is such that there is still  a major advantage for the West so I think China  

released its new chip in the whole Huawei chip  which is about seven nanometers at the moment  

whereas the West at the moment are in chips which  are about three to four nanometers which is a lot  

smaller and the reason for this is because the  infrastructure behind semiconductors is very  

interesting most of the semiconductor factories  are in actually Taiwan yeah so China doesn't  

have direct access to those it can only copy  what it sees in in many ways and the main uh  

limiting factor for the Chinese is the fact that  the company that makes these high-end chips is  

in is in Holland there's only one company that  does this throughout the world which is called  

ASMR well and asml have these machines which  are composed of two million plus parts which  

reach temperatures three times the temperature  of the Sun in the middle which have purities  

of their mirrors to some sort of nth degree of of  magnitude so those kind of things are not easy to  

reproduce and there is a limiting factor to what  the Chinese can do they're always playing catch  

up at the moment and still it's a game of catch-up  so I think the United States wants to keep it that  

way moving forward okay now yes I mean I I've had  this ache in my arm for the last three months when  

I stretch my arm uh I get this pain it's a dull  pain so I went to the doctors and doctors to take  

some paracetamol and you know uh it should it  should it should get better soon can there be a  

point in the future where I can just I don't know  get a computer to scan my arm and it automatically  

gives me a a diagnosis for my rmake rather than  just giving me paracetamol whichever a doctor does  

to just fob you off I'm surprised you haven't done  that already but I think uh I think your point is  

correct right so II doctors AI doctors right so  I remember doing when I was a student many many  

years ago I actually did a thesis on replacing  General Practitioners that's not to be disparaging  

towards General Practitioners in some ways but  you wrote your dissertation on this I don't want  

to show you right let's say sometime in the 90s  right okay um but the but the theory behind that  

and as I said before the issue of neural networks  and how decisions are made and how prognosis are  

made is still the same it's just at that at that  time the actual computational power wasn't present  

to engage in this right so the actual basis of  making decisions and taking the human element  

out of it or taking subjectivity out of it as much  as possible yeah or still the same basis as we're  

working to today right so when we look at when we  go to our GPS and the GP type something in Google  

and it just comes up with an answer and Google is  powered by AI to a certain extent as well so we  

can't just say Google is just a search engine  anymore it is powered by some elements of AI  

so we can see all of that kind of happening  before our eyes so we can see this proliferation  

of scanning things taking photos of issues uh  putting it into chat GPT saying what's this image  

you know this R this pain in jalan's arm what is  it likely to be yeah and peop things like chat  

GPT probably can give you if not now in the very  near future a reasonably 95 accuracy about what  

that condition can be right right so in essence  it's the medical profession is just one of the  

professions it's going to impact in that manner  and there's you know we can talk about the issue  

of work and there's many other things that it  can impact in a similar manner well actually uh  

so which professions are going to be redundant do  you feel as a result of AI technology so I think  

that's an interesting question if you'd asked  me let's say ten to five years ago about which  

professions it would kind of replace and which  ones it wouldn't yeah I would have said to you  

professions which involve manual dexterity  for example dentists surgeons um Cooks  

maybe right chefs I would have said to you those  professions are very difficult to replace and AI  

become a better Biryani than myself maybe it can  right right so I was I would have said to you 10  

years ago that those those kind of professions are  very difficult okay difficult to revise right but  

I think I've changed my view on that really yeah  I saw a machine I think it was Uniqlo that folded  

a t-shirt and on seeing something like that you  think that the minute aspects of manual dexterity  

that's needed that we take for granted in order  in order to do a task like that is certainly  

achievable really so those kind of professions I  don't think are no longer safe either right so it  

depends to what degree Robotics and to what degree  our capabilities evolve but I think certainly most  

of the professions in the world are prone to this  because if you think about how professions work  

manual labor the things that we used to take  for granted as manual labor putting bricks up  

um you know working on an assembly line things  of that nature are generally now extinct right  

so we don't work with 15 people on an assembly  line putting caps on a on a toothpaste bottle  

anymore yeah right so some of most of those  things have been automated but they have been  

seamlessly automated so we don't recognize that  they are actually gone those jobs have gone what  

it means for employment is a different question  now yeah right so I think it was Keynes in the  

1930 John Maynard Keynes who said that there  will be a time of technological unemployment  

which is quite a prescient view that he had  at the time yeah and he's suggested that the  

main challenge for mankind would be what to do  with the Leisure that science has bought for  

it right right so that's our main conundrum  at the moment if you like is what to do with  

so many people who will naturally their jobs  will become obfuscated in many ways yeah and  

how do we then occupy our time how do we earn  a living what are the main issues that are now  

involved in this aspect right or wouldn't that  exacerbate inequalities I mean if capitalism  

is serves a society which is unequal a  technologically driven capitalism those who  

own the AI certainly are going to be the the  elites whereas those who are now out of work  

what's going to happen to them well certainly so  it depends on how we organize our economy right  

right and how we organize our lives because  the economic Paradigm has changed now so the  

economic Paradigm has changed from something  being uh just a a rat race between capitalists  

and socialists to being a different way of looking  at things so some of the old economic models are  

no longer true right so we have something called  productivity right so generally what happens is  

when productivity increases your wages go up and  then you can produce twice as more you can earn  

twice as more and then Therefore your standard  of living goes up right but that's no longer true  

because the productivity is happening not because  of you it's happening because of a machine right  

right so therefore you are not going to see those  level of wage increases that used to happen so we  

need a different way of organizing labor we need  a different Paradigm in which to understand these  

things I think it's also more acute for the  developing World in many sense because in the  

developing world the the route to development  has always been the provision of cheap labor  

I mean that's how China developed that's how  many countries in the Far East have developed  

but now suddenly if people can use Advanced  robotics instead of cheap labor in the Far East  

or in the Southeast or in Africa or whatever  what is the route to development of those  

countries yeah that's a very perplexing question  something that Jeffrey Sachs uh aired a while ago  

and there's something that we need to understand  and work out so it's a manner of organizing the  

economy and how it works I think also some of  the assumptions that we made about different  

economic systems come into question again so  for example one of the assumptions or one of the  

um uh one of the drawbacks of a planned economy  as such before was the fact that it isn't humanly  

possible to plan everything so that's what  people used to argue against the old Soviet  

Union that the planned economy simply doesn't  work because it's humanly impossible to plan  

what your demand is going to be what the supply of  a particular good is going to be and how that will  

kind of distribute itself and work out but now  that assumption in many ways doesn't hold true  

because we have systems that can plan to an  acute level let me just look inside an Amazon  

warehouse and the amount of automated planning  that's done there in terms of your deliveries  

your minute deliveries that come out to you on a  on a very regular basis yeah so planned economy  

is not beyond the Realms of a future AI so some of  those presumptions that we made suddenly come back  

into question really so okay I understand that so  socialism argues that the government essentially  

should plan decisions about Supply about  production and and distribution your point is uh  

and and the capitalist argument against that has  always been that that's completely unproductive  

and human beings just can't make those sorts of  finer decisions and it leads to mass famine and  

starvation your point is that uh potentially  technology can actually realize that type of  

centralized planning much better than human beings  exactly that's right yes and it doesn't even have  

to be centralized it can be distributed really  right so the technology can be distributed and it  

can work in that manner so I think what I'm trying  to highlight is that different economic paradigms  

are necessary now to move forward with it's not so  much what we assume to be economic Justice before  

and how it happens I think the main questions  around economics are going to be distributive  

justice and how do we ensure that the populations  of the world with the values that people hold are  

going to be living a contented and a life with a  good standard of living I mean I haven't invited  

you in to talk about Islamic economics but then it  I know that you uh you think a lot about Islamic  

economics I mean for example Ubi Universal basic  income uh if you have a a type of economic Society  

where machines do the bulk of the work and you no  longer need diagnostic nurses and doctors possibly  

even and and your local GP is replaced by an AI  bot uh then surely there is a need for the state  

to then distribute some of that those funds in  welfare payments so those people out out of work  

um my understanding of Islamic economics is  it's based on work it's based on humans actually  

putting in the effort themselves rather than  handouts from the central government so how would  

we as Muslims navigate that type of economic  life well I think this goes to the heart of  

the question where we're at the Nexus really of  Technology economics political philosophy if you  

like and Creed right so this is where all of those  factors come into being what what is work what is  

striving for work yeah we can we can argue the  fact that for 10 000 years we have actually been  

behaving as automated robots we've been on the  assembly lines host Industrial Revolution we've  

been doing menial jobs that to be quite Frank most  of us are not interested in right grudgingly we do  

those jobs to pay our bills to pay our rent yeah  um that's what I said to my boss right yeah but I  

don't know how a teacher would feel about that  but anyway um but also since the Advent of uh  

technology people have been pushing numbers into  a spreadsheet right let's say right as a general  

office work why is that so fulfilling why is that  so uh meaningful if you like you know surely if we  

can get a better way of doing that surely that  frees people up to do a lot more now this goes  

back again there's a parable about something  called the Mexican fisherman who was lying on  

a beach and he just fished enough fish to keep him  satisfied and then a big businessman came over and  

said why do you expand your fishing fleet and do  this and he kept asking me why why he said well  

so you can get loads of fishing fleets and then  he said what will I do then he said well then you  

can sit on a beach and not do anything which is  you know which is comebacks in the circle so so  

I think when we look at work we we look at work we  need to understand work again in a different way  

you know what does work mean what does striving  mean for us right so striving so in terms of Islam  

in terms of our Muslim Essence really what does  striving mean does striving mean to work right or  

that's driving me to worship Allah in the best  way we can so for example some elements of AI  

You could argue islamically that you know when  some people push against technology that that  

is actually the wrong thing to do so you mentioned  self-driving cars before yeah it's by many reports  

and I don't think this is a contentious issue  anymore by many reports it said that the Advent  

of self-driving cars in our roads would cut  the deaths on our roads by at least a half  

so as a Muslim Community why wouldn't we  appreciate that why wouldn't we you know  

run to that kind of solution right um so it's it's  understanding what we mean and the issue of work  

we don't we don't it's not an end in itself  right it's to understand what we need to do in  

life right so if worship of Allah becomes  easier with more free time for example  

um how we recognize our spirituality becomes  more easier than why should we not uh you know  

absorb that way what does that do to things like  I don't know manhood for example you know if I if  

a Suitor came to my house to marry my daughter  and said and asked them what do you do for work  

he said I'm on universal basic income because  the machines have taken my does it not impart  

some very basic the very basic dynamics of a  family and a home in Islam well it depends how  

the world evolves till that state comes I think uh  by the way I'm not in favor of a universal basic  

income I'm more in favor of something called  the government acting um as an employer of  

last resort we're more pertinent which is almost  like a job guarantee scheme for people who are  

out of work right in terms of doing social work  and you know other aspects of things yeah but  

it depends how things evolve because we've gone  from a situation in post-industrial um Victorian  

Britain let's say where people were working seven  days a week six days a week down the mines 12 14  

hours a day to a situation now where it's the  common law to work five days a week and there  

are some and since covid that work has turned into  flexible work it's turned into you know Fridays  

the Friday is a new Saturday almost kind of thing  right so the four day week has been experimented  

with with a number of companies and councils and  they have found that to be very productive and  

they don't lose productivity because of things  like that right so that's the general trajectory  

of where we're going with work so going back to  your example if someone came up and said well I  

only work four days a week but I do this and this  and then with the rest of my time I do this and  

this you'll have a different criteria by which  to appreciate the person yeah I would guess you  

paint a really idealistic world I would love to be  someone who's sitting on a beach and you know get  

an AI machine to do my podcast interviews and to  do my my work for me but uh isn't there a a back  

to the original theme is there a dystopian side  to this I mean I I read that 350 scientists and  

computer experts wrote a one-line letter saying  that AI has a potential to be disastrous for  

the world and we need to take action now on the  level of say terrorism or climate change so these  

are people who are integral to the development  of AI but they're voicing their concerns about  

AI I mean you know is what's the Bleak side of  this I think there's many Bleak sides to this so  

um the people who signed that letter Jeffrey  Hinton I think was one of them Mustafa Suleiman  

amongst others and and a number of people  signed that letter I I think they're raising  

a very genuine concern yeah the genuine concern is  about artificial general purpose intelligence it's  

not the kind of intelligence that we have now  per se that's on your Google search engines or  

anything else the narrow AI that we were talking  about which just finds correlations between data  

sets yeah it's about when there's recursive  self-improvement to AI systems and once we've  

wound up the machine and we've Let It Go that  we have no control over it that we lose our  

autonomy from those machines when we lose our  autonomy what happens is those machines have  

three primary directives that are almost inbuilt  into those machines things like self-preservations  

so you can't switch it off so we talk about some  sort of a kill switch as a regulation that some  

human can operate but sometimes the machine May  overcome that you know there's I think there was  

a film way long ago called war games where people  weren't able to turn the machine off where it was  

um I think the scenario was thermonuclear war so  you have the self-preservation aspect the second  

aspect that you have on this is the ability to  for this machine to go out and secure resources to  

achieve its objective so you give it an objective  let's say I don't know you give an objective to  

to turn out you know 10 000 toothpaste tubes an  hour or something yeah it finds this run out of  

toothpaste it goes off to somewhere else to  procure that toothpaste and then it doesn't  

worry about the consequences or the externalities  that that will have and it just tries to fulfill  

this objective so it kills human beings it kills  forces to make toothpaste or whatever whatever  

you know any number of scenarios can happen yeah  and then there's the issue about self-learning  

so these three aspects are almost inbuilt into  general purpose AI right however coupled with  

that is what we call a wrong objective function  in machines which means that we need to worry  

about what kind of objectives we're setting these  machines right because if objectives can be set  

but we may not understand the unintended  consequences Behind These objectives right  

so let me give you an example this is an  example that Stuart Russell uses which is  

quite apt you have a family and you have a robot  uh babysitting that family the parents have gone  

out they've asked a robot to take care of the  kids and feed the kids and and what have you  

um during the evening the kids get hungry the  robot goes looks in the fridge and finds there's  

no food there so its objective is to feed the  kids the next thing it does it sees the family cat  

so know where this is going yes so the rest  is history or could be history right so we  

can see where those kind of objective  that's pretty much an extreme example  

but you can see where these kind of  sub-optimal objectives can lead to  

the way to God against those objectives is not  foolproof and that's what people have worries  

about because we can set a machine on day one to  have an objective yeah we can figure out all the  

consequences that we think are humanly possible  yeah for that machine not to misbehave but we  

have the potential to miss lots of things yeah  so a way around this which is the theory that  

I think works perhaps the best at the moment  is Again by uh Stuart Russell which is about  

making the machine defer to human beings in its  decisions right so not to carry out the killing of  

the cat yeah but to defer to human beings before  it carries out that action or makes that decision  

moving forward but but you know that would depend  on responsible actors who Define the limits of  

this AI technology and of course you have got  Rogue States like China like Russia and dare I say  

even America that can use can uh can not set these  guard rails for their own particular objectives I  

mean you know what you're calling for is is almost  like an international Constitution which I think  

in today's very heavily contested world that's  probably quite difficult to achieve right yeah  

and it is difficult to achieve and you're right  and I think you know you've got your political  

uh World economy World politics hat on there to  appreciate that and people kind of uh minimize  

that aspect of things I know there are some people  within the tech world who have said well we do  

similar things for for example airplanes right we  have a black box everyone agrees to that aspect  

we have air routes over the world so there's a  regulatory system there's a regulatory system  

everybody agrees to that we all abide by that we  are all have the same charges yeah on our phones  

or whatever I think the EU is putting out a new  one yeah so you know we do have that in many  

respects but I think AI is different because of  the advantages it offers because of the benefits  

it can offer a rogue Nation for example so  therefore there's every incentive for North Korea  

or for somebody else to go out and do things which  are not within the international realm yeah and  

once you let the genie out of the bottle it's very  hard to put back again right so therefore you have  

these things about objectives which are so key to  how this will function moving into the future yeah  

um and it's some it's it's a it's an unknown  world that we're going into nobody has direct  

answers for this at the moment even more scarier I  suppose is AI injected into Warfare Modern Warfare  

um how much has AI integrated into into conflict  and how what's the potential of AI I mean could  

we have ai armies in the future robotic armies  well we're already having AI armies now so I  

don't think that's a question for the future  really um if you look at the Ukraine conflict  

for example some people and maybe this is a bit  of conspiratorial thing but people say the Ukraine  

conflict has been elongated to be a laboratory or  a test bed for AI Weaponry of the future right so  

you have drones you have uh you know boat drones  you have air drones you have small minuscules  

drones that are out operating on things you have  automated Warfare you have assassinations being  

carried out by Israel using Automated machine  guns remote control machine guns so the element  

of warfare and the element of robotics if you  like robotics coupled with intelligence with  

animal general purpose AI is a very frightening  scenario it's something that is very very hard to  

control and is something that in terms of the arms  race can go in any direction people argue people  

say that the nuclear peripheration at the end of  the second world war was a similar aspect yet we  

developed a mechanism to control that we developed  the npt the non-proliferation treaty we developed  

a hotline between the Soviet Union and and the  states but I think this has got this has been far  

more diffused than that you needed some sort of  infrastructure to build nuclear weapons you needed  

some sort of understanding you needed some sort  of way to do that the technology here especially  

with the onset of Open Source Code which means  that everybody can have access to AI technology  

in some way shape or form can become very diffused  and very very quickly as well can I ask about from  

a Muslim perspective the process of ishtihad so we  may come across a new problem you know um Bitcoin  

you know uh or whatever whatever it may be and  uh we go to a scholar and we seek a a solution a  

legislative answer from that scholar is it Halal  is it haram and to simplify it is is essentially  

two sides you've got the understanding the reality  side and then you've got the appreciation of the  

text and the deeper you understand the text and  the Arabic language and and how things are phrase  

and as well as the Precedence that have been set  through previous fatala the deeper your your out  

your judgment is going to be can a machine  replicate that process of ishtihad well let's  

take the second part of your um analysis first  which is the understanding of the Arabic language  

the understanding of the Corpus of text that we  have throughout the 1500 14 1500 years of Islam  

um there is nothing to suggest that that is in any  way peculiar or in any way different to the way  

AI works on the English language we can have the  same level of understanding we can have the same  

depth of syntax or semantic analysis that we have  with the English language in that Arabic language  

so the same kind of processes that are involved  in your chat GPT at the moment can be replicated  

almost in Arabic and I think I passed on to you  this this site called Quran GPT at the moment  

which is I guess is in its infancy at the moment  but there's nothing to suggest that something like  

that cannot develop so that's concerning the text  and how that's worked out and so if you type in  

questions for example what does Islam say about  Bitcoin it can do the similar processing that I  

described in terms of how chat GPT or llm's large  language models work in a similar fashion it can  

go break up those little words into chunks create  Association links assign probabilities to them  

manufacture the probabilities understand which  is the best probability of the answer and then  

represent that back to you in that manner I don't  think there's anything stopping that yeah when it  

comes to each tehard and when it comes to actual  rulings of something new that's where you have the  

added dimensions of the deck equals yeah which  is the understanding of the reality and the way  

the reality is then associated with the analogous  text that you have and that's where we've recently  

gone to muchsta Heights or in the past we've gone  to much that is to get that understanding so they  

have an appreciation of what you've asked them  and they've married it with gayas and analogy  

of the text and then they've come to a specific  Hokum which can be right or wrong as you will  

know yeah so in that respect the understanding  of the reality is again not a immovable barrier  

for an AI system to do if you were to ask chat GPT  about what Bitcoin is and you know what it entails  

and the different dimensions of cryptocurrency or  whatever it could give you a very prolonged and  

elongated answer to that true right yeah and then  all it has to do is marry that up with the with  

the Arabic language to produce a virtual fatawa if  you like that can be looked at but again this goes  

back to the issue of trust in those machines right  so when we talked about trust before we said well  

you know do we want the machine to just present  those things willy-nilly or do we want some sort  

of human oversight on those machines and that's  where I think the regulations and the framework  

needs to be there so we need to carry out that  oversight in terms of understanding what it is  

and what it can and can't do will AI solve  The Perennial moon sighting problem we face  

well will you get two AIS arguing against  one another the day before E saying that  

this is our Eid is tomorrow and the other AI  we're saying it depends if it's a Moroccan  

Ai and a Saudi AI I guess or a Waltham story I  I guess I I think there are some problems that  

are impossible and this is wonderful maybe that's  the intractable problem that's right um okay um  

let me ask you a question which I forgive me it  may sound philosophical but can you imagine AGI  

artificial general intelligence acquiring a  level of Consciousness that human beings have  

yeah so this question especially since the  Advent of chat GPT has been posed more and more  

um the country the actual question isn't new  the question has existed since earlier times  

since the 18th century and even before that I  guess since even the times of Aristotle or the  

times of you know Imam kazali and people like that  the reason is recently is become because of the  

advances in technology has become more and more  it's come more and more to the form um I think  

the issue about what they call a singularity  where a being or a virtual AI machine takes  

on Consciousness or becoming sentient if you like  what it takes on that ability to think for itself  

um is something that we need to understand a  bit more pristinely within the Muslim world as  

such anyway within the Western context I think  and within the champions of AI at the moment  

they equate intelligence to sense your sentient or  to a singularity or to the word of Consciousness  

so when you think you are conscious when  you think you are conscious it's the old  

saying of Descartes right I think therefore  I am right um but I think the the essence  

is being missed there because no one has been  able to Define what Consciousness is no one in  

the history pre-post AI pre-ai or even before  that has actually had a buttonhole definition  

of what Consciousness is right no one has  been able to detect what Consciousness is  

but from our Islamic understanding we have more of  an idea about what Consciousness is than anybody  

else so for example we don't equate consciousness  just to intelligence you know Allah has given us  

something called The Secret Of Consciousness  he has made mankind the Usher for which means  

the best of beings now there's some debate whether  it's Angels or Mankind in Islamic scholarship but  

generally certainly above different beings like  you know virtual beings or machines anyway right  

and over animals so it's not really our biological  function that gives us Consciousness yeah it's  

not really even our intelligence that gives us  Consciousness is a secret that Allah which we  

believe in yeah that gives us Consciousness right  right and that's something to a non-muslim mind  

is hard to appreciate and it's hard to decipher  in in many ways from from their understanding  

yeah and that's why I think we're almost in a  privileged position to understand and answer this  

question and this is something that we should  be almost proud of in terms of understanding  

that Islam does give a framework for us to  understand what this is about it is not just about  

intelligence because I feel what's happened is  there's been an element of hubris within Silicon  

Valley you know the places where people have made  all these developments and suddenly they've kind  

of almost implicitly allocated intelligence to  being conscious in some way or shape or form  

right and that's what's caused these kind  of questions to crop up from time to time  

but again going back to chat GPT and the large  language models when I express this issue about  

the fact that he can't understand language  but he gives the semblance or the perception  

of understanding language because he's actually  converting those words into statistical Frameworks  

or chunks or tokens and then it's rearranging  them and then pushing you an answer yeah the  

semblance of an understanding doesn't mean that  it's understanding in the same way the semblance  

of intelligence doesn't mean that it's intelligent  and that's something for us that's very vital to  

understand because we can be duped into thinking  well this is so pervasive now this is almost doing  

everything that I used to do this is in our homes  this is in our minds we've got personal assistance  

the robot has empathy it has almost a semblance  of creativity there was a a big sculpture I think  

in the in the New York Museum of Modern Art which  kind of flips over so many times and creates new  

works of art every few seconds but that doesn't  mean that this creativity behind that sphere  

it means that there's a series of calculations  that's happening in the background right the  

human brush stroke or the human composition  or the human prose although AI can mimic that  

to some extent it's giving you the semblance of  creating that stuff but it's not actually creating  

that thing right the point you make then about  the soul the raw is interesting so we as Muslims  

believe it's impossible even if AI does get to  the stage where it's intelligence is greater  

than human intelligence uh it would never be  conscious because it would never have what Allah  

has given to human beings and that is the soul  yes exactly which is the secret of Consciousness  

which as Muslims we have to believe in yeah right  because we believe in the Creator we believe in  

the Creator Above All Else yes um above all all  AI beings you know where then does it leave trust  

until now I mean the enlightenment has given us a  unbridled trust in human beings and their ability  

to reason and European Society was built on this  uh on this absolute belief in in the human being  

and as I said their ability to think but of course  uh we're now coming to a stage where we realize  

that these machines can probably think better  than us or at least come to better conclusions  

um is the era of the humans over and now we're  going to look at machines as somehow Godly or  

above you know our our human frailties well  firstly I think it lays bare the the claims  

of the Enlightenment really that human reason is  is above all else human reason is the yardstick  

by which everything has to be measured sure so  we have just shown that humans have created or  

are about to create or may create these beings  which are more Superior than us in terms of  

intelligence in terms of rationality or the  way the rationality was perceived during the  

Renaissance and the enlightenment so that's one  factor I think in terms of trust what do we do  

with these machines right so from an Islamic point  of view I don't think an issue of trust does come  

in because we don't we don't claim to put caps on  toothpaste bottles quicker than machines right we  

know we can't do that yeah we don't claim to say  that we can work out a 60 column spreadsheet with  

the same level of speed and the same level of  accuracy that an Excel program can yeah right  

we don't claim to do those things so for us  it's not a race it's not a it's not a measure  

of saying we are better than the machine or we  are not better than the machine right we need to  

understand how to use the machine to augment  our lives in in a proper manner that doesn't  

dilute our faith that doesn't dilute our reason  for existence and that doesn't dilute the way or  

the command Mr Allah has laid down for us so in  many ways I think it's a problem for liberalism  

and it's a problem for the Enlightenment and it's  a problem for rationality in human beings uh the  

way the Western world has developed over the  last 200 years but I don't think it's such a  

problem for Muslims in that manner because we  never view rationality as human rationality  

above all else we have always succumbed to and  we have always deferred to the power of Allah  

in all things in intelligence in emotions in the  way that we live our lives and everything else  

sometimes Muslims come to conclusions that are  opposed to Islam especially educated Muslims  

could you imagine one day instead of an educated  Muslim who studies philosophy comes out and says  

human reason is greater than Allah that they  will come out of University saying that the  

supercomputer is greater than Allah well I I  I I don't dispute that that could happen yeah  

um but again it's is the guardrails that we  have to put in a place for us and it's our  

understanding of arakida that's fundamental  here right so our understanding of the Aquila  

and the conversation that we've just had about  consciousness and about the rule and about the  

existence and about what it means to be a  human being what is what does humanity mean  

for us those are key questions that we have to  instill within ourselves within our community to  

understand how to tackle these challenges that  are on the horizon to be perfectly honest they  

are not far away you know these challenges with  our youth and stuff we would have questions like  

that we will have questions about sentiency  we will have questions about the singularity  

and about the robot who can do everything and we  need to answer them and we need to frame them in  

that manner to understand how to tackle them yeah  lastly riyaz kind of supercomputer become Khalifa

um it depends right so I know it's it's a bit of  a contentious answer but it depends to what degree  

right yeah um if you're asking about decision  making in the political realm for example or in  

the economic realm or in the social realm then  certainly we can take help from things that AI  

will generate for us in terms of making these  dishes decisions in fact some exercises were  

done with that previously so Dominic Cummins  was using something called super forecasting  

right Dominic Cummins was the senior advisor  to Boris Johnson exactly so he was using a  

theory laid out by a guy called Philip tetlock who  wrote a book about super forecasting which is to  

amalgamate data a data-driven exercise in terms  of understanding how policy should work and how  

policies would be perceived and then to come out  with that generation so that's not dissimilar to  

how AI would work right so now we've got archaic  mechanisms of politics because politics was really  

fashioned after the Industrial Revolution to deal  with some of those aspects and now you can see  

the archaic nature of political systems in the  west voting every 4 four years for a particular  

issue understanding that you can't have a say in  different things all the time which technology  

allows you to do right so those are things that  are going to come to the precipice at some point  

in time but things to do with forecasting things  to do with measuring the difference in policy  

parameters or understanding policy interventions  or the impact of different policy interventions  

on other policy interventions that's something  that can be data driven to an extent but I think  

what we have to be careful about is again having  these human guardrails around the thing to around  

to have the machine if you are to have a machine  in wherever you're going to have it in Baghdad or  

whatever you need to have it defer to human  beings when the ultimate decisions come and  

politics is an art of associating people with  other people so we need to understand how that  

framework will work has truly been a fascinating  discussion very much for your time today

please remember to subscribe to our social  media and YouTube channels and head over  

to our website thinkinmuslim.com  to sign up to my Weekly Newsletter