The pace of developing AI that can truly engage in conversations, reasoning, and design planning like humans is accelerating, with experts predicting that AI may soon reach the level of Artificial General Intelligence (AGI). This has raised concerns among many about the potential disasters AI could bring to humanity.
Meta and OpenAI executives recently expressed their desire to create generative AI that can engage in real conversations, reasoning, and strategic planning, rather than just being proficient in single tasks or limited capabilities like current AI systems. The reason is that for AI to be considered “Artificial General Intelligence” (AGI), it needs the ability to reason and plan. AGI refers to AI with intelligence equal to or surpassing that of humans, capable of performing all intellectual tasks that a human can.
Both OpenAI and Meta have announced plans to release GPT-5 and Llama 3 modules soon, aiming to reach AGI level. Tesla’s CEO Musk stated on April 8th on the X platform, “I predict that by the end of 2025, AI will be smarter than any human, and their capabilities may surpass all humans entirely within the next 5 years (by 2029).”
Moreover, an account named Jackson on the X platform released a controversial report regarding OpenAI’s anticipated AI Q* (pronounced Q-Star) set to be unveiled in 2027. The report suggests that there is significant indirect evidence indicating that by 2027, Q* may have reached AGI level, with an intelligence equivalent to that of a human with an IQ of 145.
The exposure of Q* occurred in 2023, when online reports surfaced alleging that former employees of OpenAI had filed a whistleblower statement to the board of directors. The statement mentioned that senior management at the company had concealed the AI Q* project, which could pose a threat to human safety.
Geoffrey Hinton, known as the “Godfather of AI,” and a former Google executive aged 75, warned last September that within the next 10 years, AI technology in some aspects will surpass human intelligence, leading to catastrophic disasters for humanity. He projected that AGI would appear within 5 years, resulting in a proliferation of “killer robots,” fake news, and unemployment.
Technologies invented by Hinton have become the foundation of systems like ChatGPT and Google Bard.
On April 5th, the Adecco Group, one of the world’s largest HR services companies, released a survey of executives from 2000 large companies globally. The results indicated that approximately 41% of executives believe that employee numbers will decrease in the next 5 years, primarily due to AI replacing human workers. Some experts even suggest that AI might take over 80% of human jobs in the coming years.
Furthermore, concerns have arisen among artists and creators in the cultural and arts fields about AI replacing their creativity. The recent debut of music-generative AIs such as Suno and Udio, as well as text-to-image AI like Sora, has heightened these worries. Writer Brian Hiatt from Rolling Stone expressed surprise and unease at the songs created by Suno, describing them as alien.
Over 200 renowned artists and musicians, including Billie Eilish and Nicki Minaj, recently signed an open letter urging a ban on “predatory AI” in the music industry, fearing it could diminish their livelihoods and artistic drive. They called for restrictions on excessive AI usage and the unauthorized use of artists’ voices and images, as it infringes upon creators’ rights and disrupts the music ecosystem.
Former senior executive of market promotion at OpenAI, Zack Kass, boldly predicted in January that if current AI development trends continue, AI may represent humanity’s final technological invention, ultimately replacing roles in commerce, culture, medicine, education, and other professions.
Japanese computer engineer Kiyohara Jin shared with Epoch Times that AI advancement could intensify the already competitive creative and job markets, potentially leading to the displacement or substitution of many individuals. This presents a terrifying prospect for human development, resulting in widespread unemployment and posing numerous unresolved problems.
The rapid development of AI is also testing the capacity of many countries’ power grids, as AI advancement relies on chips, energy, and data storage. Given the crucial role of energy and networks in modern technology, national power grids may be vulnerable to mass outages due to natural disasters or cyberattacks.
While Musk believes that energy and chip issues may be resolved within the next couple of years, with companies like TSMC expanding chip production facilities in the US, Japan, and Europe, the production of more chips means a greater demand for energy.
When AI engages in machine learning, algorithms, and data storage, it demands substantial electrical support. Additionally, the development of electric vehicles also requires extensive charging infrastructure. These projects could strain the already fragile power systems, potentially threatening the essential electricity needed for people’s daily lives.
Alex de Vries, a doctoral student at Vrije Universiteit Amsterdam, calculated that by 2027, the AI sector will consume at least 85 to 134 TWh (terawatt-hours) of electricity annually, equivalent to the Netherlands’ yearly energy consumption and representing 0.5% of global electricity use. Previous reports estimated that training language models like ChatGPT’s GPT-3 could consume 1,300 MWh of electricity. This is approximately the annual energy consumption of 130 US households or the power needed for people to watch Netflix for 1.62 million hours.
However, these statistics do not account for the energy, water, and other costs associated with maintaining servers, GPUs, data storage, and answering AI queries. Scientists have found that training GPT-3 requires 700,000 liters of clean water, with answering 20 to 50 simple questions consuming 500 milliliters of clean water. OpenAI’s CEO Sam Altman mentioned that ChatGPT answers questions for 100 million people each day. If over a billion ChatGPT users ask GPT-4 questions, estimating the water usage becomes even more challenging.
Furthermore, companies like OpenAI, Meta, and Google have announced plans to release even more powerful AI models, indicating that these AIs will require more energy, chips, and clean water. This could impact the electricity and water sources essential for human survival.
Although companies like Microsoft and OpenAI proposed promoting small nuclear power plants to supply energy to AI computing and data storage centers while providing some electricity to households, they haven’t provided specific solutions regarding the locations and disposal of nuclear waste from these power plants.