Generative AI, such as ChatGPT, has become increasingly prevalent in our society since OpenAI unleashed it in 2022. These technologies have infiltrated various aspects of our lives, from our email inboxes to our medical records, sparking concerns about their societal impact. A recent Pew survey revealed that 50 percent of Americans are more concerned than excited about the growing use of AI in their daily lives, a significant increase from previous years.
Karen Hao, author of the book “Empire of AI: Dreams and Nightmares in Sam Altman’s OpenAI,” delves into the reasons behind this apprehension. In a conversation with Bri Kane, associate books editor at Scientific American, Hao sheds light on the complex and often overlooked issues surrounding AI technology.
Hao’s unique perspective on AI stems from her early immersion in the field as a reporter for MIT Technology Review. By engaging with seasoned AI researchers and academics, she gained valuable insights into the evolution and limitations of AI technologies. This foundation of knowledge allowed her to approach the subject with a clear-eyed perspective, free from the hype and sensationalism that often cloud discussions about AI.
In her book, Hao argues that companies like OpenAI can be likened to empires, wielding immense power and influence over various aspects of society. She draws parallels between the tactics employed by these tech giants and the historical patterns of colonialism and imperialism. Hao emphasizes the extraction of resources, both physical and digital, from individuals and communities, highlighting the unequal distribution of wealth and power in the AI industry.
Furthermore, Hao explores the ideological underpinnings of the AI industry, noting the pursuit of artificial general intelligence (AGI) as a quasi-religious quest for technological transcendence. This blend of profit motives and ideological fervor mirrors the fusion of capitalism and ideology seen in historical empires. By framing the discussion in terms of colonialism, Hao challenges readers to consider the broader societal implications of AI technology and its impact on power dynamics.
Through her research and reporting, Karen Hao provides a thought-provoking analysis of the AI industry, urging readers to critically examine the ethical and social implications of these rapidly evolving technologies. By adopting a historical lens of empire and colonialism, she invites us to question the prevailing narratives surrounding AI and to consider the broader implications for society as a whole.
As I delved deeper into the conversation about AI and its environmental impacts, I realized that this was the frame to also understand OpenAI, ChatGPT, and where we currently stand in this particular moment with AI. The scale of what AI is capable of now and the desired continued growth that companies like OpenAI are planning for in the near future is truly staggering.
One of the key issues that my book touches on is the environmental impact of AI, which is often overlooked in conversations about the technology. The AI industry’s path of development is exacerbating several intersecting crises, one of which is the energy crisis. Sam Altman, the CEO of OpenAI, recently announced a target of building 250 gigawatts of data-center capacity by 2033, which would cost around $10 trillion. The primary energy sources to power this infrastructure would likely be fossil fuels, as renewable energy is not currently able to meet the demands of running these facilities 24/7.
Business Insider’s investigation revealed that utilities are extending the lives of natural gas and coal plants to meet the demand from data centers, resulting in increased emissions and air pollution in communities. In places like Memphis, Tennessee, where supercomputers are being run on unlicensed methane gas turbines, toxic pollutants are being pumped into the air, further exacerbating health risks for residents.
Another major issue is the freshwater consumption of these facilities. Most data centers are cooled with water, and two-thirds of new facilities are being built in water-scarce areas, leading to competition with local communities for vital resources. This heavy industry is not only toxic to the environment but also poses a threat to public health worldwide.
Some may argue that AI will eventually solve the environmental impact it creates by finding solutions to climate change. However, the harms of AI on the environment are happening now, and waiting for a speculative possibility of AI solving everything in the future is not a viable solution. The idea of an artificial general intelligence emerging and solving all our problems is highly unlikely, as most AI researchers do not believe this is a realistic outcome.
Sam Altman, as a central figure in OpenAI, is portrayed in my book as a master manipulator who tells people what they want to hear, rather than objective truths. His ambitious goals for OpenAI and the AI industry as a whole raise concerns about the long-term impacts on the environment and public health. It is essential to consider the environmental consequences of AI development and take action to mitigate these impacts before irreparable damage is done.
In recent years, there has been a lot of buzz surrounding OpenAI and its co-founder, Sam Altman. Some have questioned whether Altman is lying or exaggerating about the capabilities of AI and its potential future. Hao, a journalist who has delved deep into the world of AI and OpenAI, sheds some light on this complex issue.
Initially skeptical of OpenAI’s claims about achieving Artificial General Intelligence (AGI), Hao discovered that there are genuine believers within the organization and the broader AI community. Some view AGI as a savior that could solve humanity’s problems, while others fear it could lead to catastrophic consequences. It’s unclear whether Altman himself truly believes in these grand claims or if he is simply leveraging them to secure more resources and capital for OpenAI.
Hao emphasizes that the focus should not solely be on individuals like Altman, but rather on the power structure that allows individuals to wield such influence over billions of lives. Altman’s storytelling abilities and persuasive nature have enabled him to garner significant power for himself and OpenAI. However, Hao argues that this concentration of power is problematic and calls for accountability and a dismantling of the current power structure.
During his investigative journey, Hao traveled to Kenya to meet with workers contracted by OpenAI for content moderation tasks. These workers were exposed to disturbing and psychologically damaging content, leading to mental health issues and societal breakdowns. Similar challenges were faced by workers involved in reinforcement learning from human feedback, where individuals had to teach AI models how to interact with humans.
One such worker, Winnie, shared her experiences working for a platform called Remotasks, which supports companies like Scale AI in their AI development efforts. These stories highlight the human cost of AI development and raise important ethical questions about the treatment of workers in this rapidly evolving industry.
In conclusion, the debate over Sam Altman and OpenAI’s claims about AI capabilities is multifaceted and requires a deeper examination of the power dynamics at play. While Altman’s charisma and storytelling skills have propelled OpenAI to the forefront of the AI industry, it is essential to prioritize accountability and ethical considerations to ensure that advancements in AI technology benefit society as a whole. The story of Winnie, a remote worker in the AI industry, sheds light on the exploitative conditions that many workers in this field face. Winnie, a mother trying to provide for her children, spent her days waiting for tasks to arrive on the Remotasks platform, never knowing who she was working for or when the tasks would come. Despite the non-traumatic nature of the work itself, Winnie found herself working long hours, sometimes up to 22 hours straight, in order to earn just a couple of dollars a day.
The lack of economic value that Winnie and other workers generate for these AI companies is a stark reality of the industry. The story of Winnie highlights the need for more ethical practices in the AI industry, particularly when it comes to the treatment of workers. Hao, an advocate for worker rights in the AI industry, believes that change will only come through external pressure and public backlash.
Hao draws parallels to the fashion industry, where international labor standards were implemented after devastating labor accidents. She sees a pathway for similar change in the AI industry, where public pressure can force companies to adopt higher standards and guarantee labor rights for workers. By holding these companies accountable and demanding better treatment of workers, Hao believes that progress can be made, even in the absence of government regulation.
The conversation between Kane and Hao also touches on the role of regulators in the AI industry. Hao acknowledges the current crisis of leadership in the U.S. and other countries when it comes to regulating AI development. However, she emphasizes the importance of bottom-up governance, with the public playing an active role in calling out companies and holding them accountable.
Despite the challenges and concerns surrounding AI development, Hao remains optimistic about the future of AI. She believes in a vision where AI systems support humans rather than replace them. By focusing on developing AI models that enhance human capabilities, rather than taking agency away from people, Hao sees a hopeful future for the industry.
Overall, Winnie’s story serves as a powerful reminder of the human cost behind the AI industry. It highlights the need for ethical practices, worker rights, and public accountability in shaping the future of AI development. Through advocacy and activism, Hao and others are working towards a more equitable and sustainable AI industry that prioritizes the well-being of workers like Winnie. Specialized AI models are the future of solving specific challenges that society faces. While the concept of Artificial General Intelligence (AGI) may not be on the horizon for solving complex issues like climate change, organizations like Climate Change AI are paving the way for leveraging AI technologies in targeted ways to address well-scoped challenges within climate mitigation efforts.
These specialized machine-learning tools are not related to large language models or theoretical AGI. Instead, they focus on tasks such as maximizing renewable energy production, minimizing resource consumption in buildings and cities, optimizing supply chains, and improving the accuracy of extreme weather forecasts.
One notable example is DeepMind’s AlphaFold, a specialized deep-learning tool that accurately predicts protein-folding structures from amino acid sequences. This tool has been instrumental in understanding human diseases and accelerating drug discovery, earning its developers the Nobel Prize in Chemistry last year.
It is essential to invest in building more specialized AI systems like AlphaFold and climate-change-mitigation AI tools. These tools can be tailored to specific communities, respecting their culture, language, and history, rather than imposing a one-size-fits-all solution on everyone.
By focusing on developing these specialized AI systems, we can create a more beautiful AI future while holding companies accountable for their actions. This approach allows us to imagine new possibilities and work towards building a more inclusive and diverse technological landscape.
In conclusion, specialized AI models offer a promising path towards addressing societal challenges in a targeted and effective manner. By investing in these tools and fostering innovation, we can pave the way for a brighter future driven by ethical and culturally sensitive AI solutions.
