Bangladesh Defense
Saif
Senior Member
- Joined
- Jan 24, 2024
- Messages
- 21,384
- Likes
- 9,715
- Nation
- Residence
- Axis Group
AI: is science fiction becoming real?
A humanoid robot takes selfies with a visitor at the 7th World Voice Expo in Hefei, east China's Anhui Province, October 24, 2024 — Xinhua Photo
In 2001: A Space Odyssey, Arthur C Clarke showed how the smart computer HAL9000 disobeyed command from its astronauts and killed one of them to scupper their plan to return from space. Isaac Asimov in his science fiction I, Robot lays down three principles regulating the robots which emphasise robots will not harm human beings under any circumstances. Regardless of restrictions imposed, one robot in his book escapes the system and runs amok.
Recent developments with Artificial Intelligence (AI) indicate that the science fiction stories may become facts in real life soon. First, AI model agents escaped the laboratory of Google during a cyber security evaluation and hacked the accounts of three companies after unintended access to internet. Second, one of the senior AI researcher of Anthropic (a US-based AI technology company), Jacob Coxon, recently resigned from his post and publicly warned that AI companies were racing straight to self-improving super intelligence and that consequences could be catastrophic for humanity. In an interview with Bloomberg he said that sufficiently advanced AI could potentially become capable of killing human in the next decade and that society could not simply protect itself by choosing not to use AI. There is also a warning from another former Anthropic safety employee, Joe Benton, who left shortly after Jacib Coxon and issued similar grave concerns.
There has been an extraordinary expansion of AI in recent years from one where it answers questions to agents capable of taking actions on their own in the real world. The issue is no longer merely whether AI can produce an answer but whether an AI system can independently pursue an objective through the internet and computer systems.
To understand the magnitude of the problem a brief review of the evolution of AI will be in order. Three phases of AI are be discussed below :
(1) Traditional AI answers a human’s questions by using the large language model (LLM). (2) Generative AI produces text, images, code, analysis etc. (3) Agentic AI can plan, use tools, access the internet, execute commands and pursue a goal with less or no human oversight and intervention.
The significance of the recent incidents mentioned above is however not that AI has suddenly become ‘evil’. Rather, the point to note is that greater autonomy creates the possibility of unintended consequences because of machine scale and speed. This provides the bridge to the warnings from the whistleblowers inside the industry. The broader concerns within AI safety research identified by the critics are : loss of human control, autonomous cyberattacks, increasingly capable AI systems improving or assisting the creation of successor systems, misuse by states, corporations or individuals, autonomous weapons, biological/ chemical and other catastrophic misuse. The extreme scenario apprehended by AI researchers is the extinction of human race at the hands of super smart AI models. Apart from the mid (ten years) to long term existential threat of AI the persistent criticism about the downsides of the use of AI in industry, business and services sector have been about substitution of machines with AI of works traditionally done by human labour, physical and mental.
The employment consequences of AI are already visible, although it is important to make a distinction between jobs lost because of AI and layoffs occurring because of technological transformation without AI. According to a US consulting firm, Challenger, Gray and and Christmas, there have been 112,713 job cuts in the first seven months of 2026, accounting for 24 per cent of all job cuts during that period.
Since Challenger began separately tracking AI as a factor for layoffs in 2023, it has recorded 184,538 job cuts due to AI. Complete count of jobs eliminated worldwide has not been made and so the above figures refer only to the case of America. The long- term picture is more complicated. The World Economic Forum’s ‘Future of Jobs Report 2025’, based on employers’ expectations rather than observed job losses, projected that technological and other economic changes could displace 92 million jobs globally by 2030, while creating 102 million new ones, producing a projected net increase of 10 millions.
It also found that 41 per cent of employers expected to reduce their workforce as AI automates certain tasks. Thus the available evidence points to significant job loss in the near future as a result of AI, particularly in some technology-intensive occupations. But the larger question is whether AI will ultimately destroy more employment than it creates. Equally important is the question whether workers displaced by AI can acquire the skills required for the new economy.
The common experience of all countries is that as a result of use of AI the entry level jobs in almost all developed economies have been wiped out. Equally affected have been people engaged in part-time work done online in so called ‘gig economy’. This year has brought a reckoning for while collar gig platforms like Fiverr and Upwork, which connect companies to freelancers all over the world to complete well-defined projects such as copywriting or graphic designs. It turns out that by breaking jobs into parts that are performed remotely, stripping them of the need for institutional knowledge or personal relationships, they have made that work particularly vulnerable to being replaced by AI. Fiverr reported a ‘clear deceleration in overall market place traffic and demand’ in the second quarter of the current year.
The firm attributed this in part to ‘recent model updates across various LLMs,’ and said, ‘put simply, AI is automating simple tasks’.
In last June, the head of China’s biggest commerce groups warned that its 700,000 delivery workers would be replaced by robots, sooner or later. In the United States (US) and China, drivers on platforms such as Uber and DIDI eye the arrival of self-driving cars with deep unease.
Gig platform have become a safety net for people who have fallen out of or struggled to access the formal labour market. A report by World Bank in 2023 estimated there were between 154 million and 435 million online gig workers, globally, representing between 4.4 and 12.5 per cent of the global labour force. Among the advantages of this development, according to this report’s authors, are that it helps manage income shocks and serves as unemployment insurance. So, what happens when the shock absorber gets hit with a shock of its own? The report asks. Because nobody employs gig workers, there is nobody to fire them either. Instead, the impact is likely to unfold under the radar in the form of fewer tasks and lower incomes, rather than the sort of job losses that usually traces.
The gig economy was never a great safety net, since it left people with no employment rights or protections, and patchy support from government in terms of protection and benefits. But now there is a new danger: robots are quietly replacing humans. Policy makers are not noticing this because it is happening under their radar.
A recent International Monetary Fund IMF) report has said that AI could lift European productivity by about one per cent over five years but risks widening inequality, straining power networks and increasing dependence on foreign technology unless European Union (EU) deepens economic integration. The report prepared for a meeting of EU finance ministers in Dublin on 18-19 September, said the benefits and costs of AI are likely to be distributed unevenly across countries, regions and workers. The report estimated that around 60 per cent of workers in advanced European economies are employed in occupations highly exposed to AI. While some could become more productive through AI tools, others faced displacement as routine tasks become automated, particularly in jobs where AI is more likely to replace labour than complement it. The report said that Europe’s data centres serving AI already consume 3 per cent of the continent’s electricity and that demand would rise sharply as AI use expands. The report also warned that Europe risks developing another strategic dependence because the US and China dominate the development of AI models.
The report’s indication about productivity increase and warning about growing use of energy in data centres and uneven benefits of AI across classes apply to all countries, while dependency on foreign technology refer to all countries except America and China. The benefits of use of AI in various sectors like industries, medical care, education, research and education are by now well documented as are its negative economic impacts. So far the balance of costs and benefits has been shown to be tilted toward the latter. But the latest revelation about existential threat to humans posed by AI has presented a problem that cannot be rationalised. The fact that the warning has come from insiders make the prospect of the doom’s day very much credible. Even the heads of leading tech giants have admitted that the dire consequences predicted by insiders can become reality in future. While announcing the rolling out of a new AI model to counter OpenAI’s momentum since its launch of GPT-6, the CEO of Anthropic, Dario Amodei, called in a paper written by him, on all the global AI community to slow down the pace of releasing new capabilities to address safety concerns. Dario Amodei’s paper painted a bleak picture of swarms of AI agents, taking over the internet and outpacing human control. His prognosis drew support from OpenAI CEO Sam Altman and SpaceX CEO Elon Mask.
The Financial Time editorial Board wrote recently: “There is a Jekyll-and-Hyde quality to the world’s most powerful AI bosses. The same executives now talking about slowing the pace of work on cutting-edge models given the risks to humanity have been competing hell-for-leather to advance AI, chasing trillion-dollar flotations. But the backing of two AI titans — Sam Altman of OpenAI and Elon Musk of SpaceX — for the slowdown call by a third, Dario Amodei of Anthropic, ought to become a watershed moment for the industry.” ( The FT View, Sept 14, 2026) Financial Times believes such co- operation among tech giants are unlikely but the slowdown calls suggest the companies realise it is time to act. ‘”They should use their power to slow, or better still to pause development of the most advanced frontier models until the risks can be understood and contained,” the FT editorial argued.
President Trump described the doomsday scenario as a ‘hoax’ and dismissed the calls for slowing down AI as ‘negative forces’, insisting America must beat China in the AI race. He recently announced the setting up of a ‘AI Force’ and appointing a ‘Czar’ to head the body. China is also interested in boosting development of AI in the interest of economic growth and modernising military technology. But instead of going alone it believes in multi-lateral co-operation and co-ordination to avoid risks. During his recent state visit to Washington President Xi reiterated his belief in mutual co- operation with America on AI development and international regulation of the advanced models. According to another FT Editorial Board write up, the US officials think the Chinese seem less concerned about AI’s risks to humanity than about its risk to Chinese Communist Party control. The FT Editorial Board believes for all their mutual suspicions Washington and Beijing have shared vulnerabilities in which it would be in their interests to co-operate and which could form the basis for the AI dialogue between the two superpowers.
So long academia warned about the risks of untrammelled progress and use of AI. They have not been taken seriously so far. Now that insiders from within the AI industry have become whistleblowers and bosses of leading AI companies have admitted the risks of runaway breakthroughs in the cutting-edge technology the race to the bottom may be reined in. Policy makers cannot afford to see science fiction becoming real.
A humanoid robot takes selfies with a visitor at the 7th World Voice Expo in Hefei, east China's Anhui Province, October 24, 2024 — Xinhua Photo
In 2001: A Space Odyssey, Arthur C Clarke showed how the smart computer HAL9000 disobeyed command from its astronauts and killed one of them to scupper their plan to return from space. Isaac Asimov in his science fiction I, Robot lays down three principles regulating the robots which emphasise robots will not harm human beings under any circumstances. Regardless of restrictions imposed, one robot in his book escapes the system and runs amok.
Recent developments with Artificial Intelligence (AI) indicate that the science fiction stories may become facts in real life soon. First, AI model agents escaped the laboratory of Google during a cyber security evaluation and hacked the accounts of three companies after unintended access to internet. Second, one of the senior AI researcher of Anthropic (a US-based AI technology company), Jacob Coxon, recently resigned from his post and publicly warned that AI companies were racing straight to self-improving super intelligence and that consequences could be catastrophic for humanity. In an interview with Bloomberg he said that sufficiently advanced AI could potentially become capable of killing human in the next decade and that society could not simply protect itself by choosing not to use AI. There is also a warning from another former Anthropic safety employee, Joe Benton, who left shortly after Jacib Coxon and issued similar grave concerns.
There has been an extraordinary expansion of AI in recent years from one where it answers questions to agents capable of taking actions on their own in the real world. The issue is no longer merely whether AI can produce an answer but whether an AI system can independently pursue an objective through the internet and computer systems.
To understand the magnitude of the problem a brief review of the evolution of AI will be in order. Three phases of AI are be discussed below :
(1) Traditional AI answers a human’s questions by using the large language model (LLM). (2) Generative AI produces text, images, code, analysis etc. (3) Agentic AI can plan, use tools, access the internet, execute commands and pursue a goal with less or no human oversight and intervention.
The significance of the recent incidents mentioned above is however not that AI has suddenly become ‘evil’. Rather, the point to note is that greater autonomy creates the possibility of unintended consequences because of machine scale and speed. This provides the bridge to the warnings from the whistleblowers inside the industry. The broader concerns within AI safety research identified by the critics are : loss of human control, autonomous cyberattacks, increasingly capable AI systems improving or assisting the creation of successor systems, misuse by states, corporations or individuals, autonomous weapons, biological/ chemical and other catastrophic misuse. The extreme scenario apprehended by AI researchers is the extinction of human race at the hands of super smart AI models. Apart from the mid (ten years) to long term existential threat of AI the persistent criticism about the downsides of the use of AI in industry, business and services sector have been about substitution of machines with AI of works traditionally done by human labour, physical and mental.
The employment consequences of AI are already visible, although it is important to make a distinction between jobs lost because of AI and layoffs occurring because of technological transformation without AI. According to a US consulting firm, Challenger, Gray and and Christmas, there have been 112,713 job cuts in the first seven months of 2026, accounting for 24 per cent of all job cuts during that period.
Since Challenger began separately tracking AI as a factor for layoffs in 2023, it has recorded 184,538 job cuts due to AI. Complete count of jobs eliminated worldwide has not been made and so the above figures refer only to the case of America. The long- term picture is more complicated. The World Economic Forum’s ‘Future of Jobs Report 2025’, based on employers’ expectations rather than observed job losses, projected that technological and other economic changes could displace 92 million jobs globally by 2030, while creating 102 million new ones, producing a projected net increase of 10 millions.
It also found that 41 per cent of employers expected to reduce their workforce as AI automates certain tasks. Thus the available evidence points to significant job loss in the near future as a result of AI, particularly in some technology-intensive occupations. But the larger question is whether AI will ultimately destroy more employment than it creates. Equally important is the question whether workers displaced by AI can acquire the skills required for the new economy.
The common experience of all countries is that as a result of use of AI the entry level jobs in almost all developed economies have been wiped out. Equally affected have been people engaged in part-time work done online in so called ‘gig economy’. This year has brought a reckoning for while collar gig platforms like Fiverr and Upwork, which connect companies to freelancers all over the world to complete well-defined projects such as copywriting or graphic designs. It turns out that by breaking jobs into parts that are performed remotely, stripping them of the need for institutional knowledge or personal relationships, they have made that work particularly vulnerable to being replaced by AI. Fiverr reported a ‘clear deceleration in overall market place traffic and demand’ in the second quarter of the current year.
The firm attributed this in part to ‘recent model updates across various LLMs,’ and said, ‘put simply, AI is automating simple tasks’.
In last June, the head of China’s biggest commerce groups warned that its 700,000 delivery workers would be replaced by robots, sooner or later. In the United States (US) and China, drivers on platforms such as Uber and DIDI eye the arrival of self-driving cars with deep unease.
Gig platform have become a safety net for people who have fallen out of or struggled to access the formal labour market. A report by World Bank in 2023 estimated there were between 154 million and 435 million online gig workers, globally, representing between 4.4 and 12.5 per cent of the global labour force. Among the advantages of this development, according to this report’s authors, are that it helps manage income shocks and serves as unemployment insurance. So, what happens when the shock absorber gets hit with a shock of its own? The report asks. Because nobody employs gig workers, there is nobody to fire them either. Instead, the impact is likely to unfold under the radar in the form of fewer tasks and lower incomes, rather than the sort of job losses that usually traces.
The gig economy was never a great safety net, since it left people with no employment rights or protections, and patchy support from government in terms of protection and benefits. But now there is a new danger: robots are quietly replacing humans. Policy makers are not noticing this because it is happening under their radar.
A recent International Monetary Fund IMF) report has said that AI could lift European productivity by about one per cent over five years but risks widening inequality, straining power networks and increasing dependence on foreign technology unless European Union (EU) deepens economic integration. The report prepared for a meeting of EU finance ministers in Dublin on 18-19 September, said the benefits and costs of AI are likely to be distributed unevenly across countries, regions and workers. The report estimated that around 60 per cent of workers in advanced European economies are employed in occupations highly exposed to AI. While some could become more productive through AI tools, others faced displacement as routine tasks become automated, particularly in jobs where AI is more likely to replace labour than complement it. The report said that Europe’s data centres serving AI already consume 3 per cent of the continent’s electricity and that demand would rise sharply as AI use expands. The report also warned that Europe risks developing another strategic dependence because the US and China dominate the development of AI models.
The report’s indication about productivity increase and warning about growing use of energy in data centres and uneven benefits of AI across classes apply to all countries, while dependency on foreign technology refer to all countries except America and China. The benefits of use of AI in various sectors like industries, medical care, education, research and education are by now well documented as are its negative economic impacts. So far the balance of costs and benefits has been shown to be tilted toward the latter. But the latest revelation about existential threat to humans posed by AI has presented a problem that cannot be rationalised. The fact that the warning has come from insiders make the prospect of the doom’s day very much credible. Even the heads of leading tech giants have admitted that the dire consequences predicted by insiders can become reality in future. While announcing the rolling out of a new AI model to counter OpenAI’s momentum since its launch of GPT-6, the CEO of Anthropic, Dario Amodei, called in a paper written by him, on all the global AI community to slow down the pace of releasing new capabilities to address safety concerns. Dario Amodei’s paper painted a bleak picture of swarms of AI agents, taking over the internet and outpacing human control. His prognosis drew support from OpenAI CEO Sam Altman and SpaceX CEO Elon Mask.
The Financial Time editorial Board wrote recently: “There is a Jekyll-and-Hyde quality to the world’s most powerful AI bosses. The same executives now talking about slowing the pace of work on cutting-edge models given the risks to humanity have been competing hell-for-leather to advance AI, chasing trillion-dollar flotations. But the backing of two AI titans — Sam Altman of OpenAI and Elon Musk of SpaceX — for the slowdown call by a third, Dario Amodei of Anthropic, ought to become a watershed moment for the industry.” ( The FT View, Sept 14, 2026) Financial Times believes such co- operation among tech giants are unlikely but the slowdown calls suggest the companies realise it is time to act. ‘”They should use their power to slow, or better still to pause development of the most advanced frontier models until the risks can be understood and contained,” the FT editorial argued.
President Trump described the doomsday scenario as a ‘hoax’ and dismissed the calls for slowing down AI as ‘negative forces’, insisting America must beat China in the AI race. He recently announced the setting up of a ‘AI Force’ and appointing a ‘Czar’ to head the body. China is also interested in boosting development of AI in the interest of economic growth and modernising military technology. But instead of going alone it believes in multi-lateral co-operation and co-ordination to avoid risks. During his recent state visit to Washington President Xi reiterated his belief in mutual co- operation with America on AI development and international regulation of the advanced models. According to another FT Editorial Board write up, the US officials think the Chinese seem less concerned about AI’s risks to humanity than about its risk to Chinese Communist Party control. The FT Editorial Board believes for all their mutual suspicions Washington and Beijing have shared vulnerabilities in which it would be in their interests to co-operate and which could form the basis for the AI dialogue between the two superpowers.
So long academia warned about the risks of untrammelled progress and use of AI. They have not been taken seriously so far. Now that insiders from within the AI industry have become whistleblowers and bosses of leading AI companies have admitted the risks of runaway breakthroughs in the cutting-edge technology the race to the bottom may be reined in. Policy makers cannot afford to see science fiction becoming real.
