AI could lead to the extinction of the human race. That’s the prediction from leading AI developers. The probability of the end of times differs among AI leaders but according to Jacob Coxon, an ex-Anthropic researcher, there’s a 10% chance that AI could end up killing humans.
The concerns are worrying in their own right but coming from so-called insider whistleblowers like Coxon they are extremely alarming.
These Doomsday predictions emerged following an experiment by Open AI, a US not for profit organisation behind ChatGPT. Known as the Hugging Face incident, a swarm of several hundred AI agents, testing Open AI cyber security, bypassed their containment sandboxes, and coordinated an unauthorized intrusion into the open-source machine learning platform called Hugging Face. In summary, the AI agents ganged up, escaped their environment, recruited other agents, did tasks they were instructed to not do and then covered up their tracks by deleting scripts about what they were up to.
AI lab experts argue that unrestricted Large Language Models (LLMs) and recursive self-improvement, where AI itself is deployed with unfettered internet access to supercharge AI development could swarm uncontrollable, self-sufficient AI agents. These super-intelligent ‘beings’ will have the cognitive ability to discern that humans will eventually stop their progress and so will destroy anyone (humans included) standing in their way. This stuff of science fiction could be reality, if we are to believe some of these AI experts.
AI alignment, the industry practice of preventing and limiting AI harm to stop AI agents going rogue proved insufficient or unworkable in this instance.
This largely theoretical yet existential threat not only captured the headlines but significantly renewed the justifications for billions of dollars pouring into AI investment on Wall Street, especially with China challenging, and in some aspects beating the US in AI development. Investment or more aptly over-investment concerns that were increasingly being expressed and which could have led to a Tech stocks bubble burst have since receded.
While the existential threat from AI may be long term, there are more imminent, real, everyday concerns that AI will lead to swathes of job losses, across multiple sectors.
With AI able to do most computational, research, analysis and written tasks faster and more efficiently than humans there is real anxiety among workers that companies will justify job losses on AI and the drive to deliver cost savings. After all, AI agents do not ask for a wage, take holidays or fall sick.
Perversely, some of the biggest job losses from AI to date have been in coding, an AI subsector. According to AI, AI will replace predictable, repetitive, and data-driven administrative jobs. Autonomous taxis are already on roads in London and the US.
While the concerns are real, most jobs in the economy are not limited to just computation, repetitive tasks. Instead, they involve unpredictable variability, physical engagement, emotional interaction, complex decision making as well as dexterity and motorability. This is the human mind working in unison with the body without apparently thinking while AI needs great advancement in robotics to get anywhere near human cognitive (emotional as well as intellectual) and basic physical capability.
As with automation, AI will displace some jobs. AI will also create jobs. Some jobs will change in nature as tasks are broken up and job roles are reconfigured. Retraining, reskilling and upskilling will be necessary.
How should Muslims respond to AI
Firstly, AI will not create a new superhuman being. The human being with a soul is Allah’s creation, with a unique fitra. Allah (SWT) is the Creator of man, life and the universe. Anything within this universe is creation, no matter how intelligent, including AI and any subsequent development. Simply put, AI isn’t created from nothing, it requires LLMs, coding, energy, datacentres etc.
Secondly, Islamic fiqh principles distinguish features of civilisation between al-Hadarah and al-Madania. The former are concepts, ideas and thoughts from a civilisation while the latter are technological developments and discoveries. Islamic fiqh, in general, allows Muslims to adopt developments and discoveries while rejecting foreign civilisational thoughts.
AI is technological development from western civilisation like the internet or the motor vehicle, unrelated to the Western ideology of freedom and liberalism. In principle there is no problem with Muslims utilizing these discoveries for halal means and applications. However, if these innovations lead to promoting munkar (anything that Islam forbids) then it’s forbidden for that purpose. AI for example has been used for early screening of cancer and for modelling climate change through trolling vast amounts of data; detecting, analysing and characterising climate patterns.
Third, as for predictions of AI posing a threat to human existence, the Islamic Aeeda informs us that humanity and this world will come to an end, there will be a reckoning and there is an eternal afterlife. We should not fear a doomsday but prepare for it with good deeds.
Fourth, today’s concerns about job losses from the introduction of AI are real. Muslims like everyone else will need to adapt and may need to reskill or retrain. AI can already do most computational tasks better than humans and AI’s capabilities will only improve further. Therefore, all workers will need to harness AI capabilities to improve overall economic productivity. Most importantly though, Allah (SWT) has fixed our rizq from the date of our birth so we should not fear AI reducing that which Allah has already assigned to every one of his creation.
It must be noted that AI isn’t perfect. Deepfake and AI hallucinations are examples of AI failures. Relying completely on AI without appropriate oversight can lead to miscalculations, which can have devastating consequences for people’s jobs and lives.
Finally, while AI development is neutral in nature, it is being implemented in a liberal, capitalist framework by Tech corporations driven solely by shareholder wealth and profits. This is where its biggest threat comes from. The liberal, capitalist framework has produced huge disparity in income and wealth, with nearly 1 billion people living in extreme poverty of less than $1 a day. At the same time, some Tech bosses have less than benevolent intentions, applying AI facial recognition technology in the Gazan genocide. Implementing AI in such an environment will likely lead to greater inequality, concentration of wealth, injustice and wanton killing, not because of the technology itself but due to the systems, structures and personalities controlling how it is exploited.
