The internet is full of AI enthusiasts who rave about the opportunities provided by AI-driven tools like Chat GPT. The use of AI is growing exponentially. You will find AI integrated into most of the everyday IT tools we use. AI is transforming our world, but at what cost? This month, the UN released a groundbreaking report: Governing AI for Humanity, which addresses this question. In this post, I’ll ignore the opportunities for now, and explore some of the risks that were raised in this report.
Top of the list of worries is the risk of damage to information integrity. This refers to the spread of misinformation, or the risk of people impersonating others using AI-generated, deep-fake content. Everybody can think of good examples of that. That Viral image of Pope Francis wearing a white puffer coat. In addition to this, media manipulation and the flood of fake news stories threaten to distort public opinion, undermine trust in elections, and discredit democratic institutions. The regulations to combat disinformation are always one step behind technology. Within a few years, Generative AI has become a big part of our world, becoming a fact on the ground and inseparable from our work processes. It is proceeding much faster than the legal framework, which is desperately trying to keep up with it.
Next is the intentional use of AI in armed conflict by state actors, which is the sinister use of autonomous weapons systems in modern conflict. Yes, Robocop! AI is being integrated into military technologies, from autonomous drones to complete surveillance and targeting systems without any human oversight. These systems analyze vast data sets, examine patterns of behavior, and then make life-and-death targeting decisions. This trend raises profound ethical and legal concerns, as it introduces new dimensions to modern warfare that threatens our security and human rights.
Another big concern mentioned in the report is inequalities arising from differential control and ownership over AI technologies. In other words, it allows a few big corporations to dominate these technologies and concentrate huge amounts of wealth and power. These companies have access to most of the world’s computational resources and data that is needed to create AI models. This concentration of ownership allows them to shape the AI landscape, control key innovations, and of course, to reap the gains. It fuels the digital divide, the gap between those who have access to modern technology and those who do not. This could evolve into an “AI divide,” where entire regions or communities are excluded from the benefits of AI advancement, further widening the gap between the rich and the poor, developed countries, and under-developed countries.
A related topic is the problem of discrimination, particularly against marginalized communities. In other words, this is the danger of the inbuilt bias found in AI systems. AI systems are only as good as the data they are trained on. The problem is that the data reflects existing societal biases. AI can unintentionally perpetuate or amplify those biases. A 2022 report by UNESCO warned of the effects of AI on women in work. AI technologies which are often used in recruitment can introduce biases. They can unintentionally favor male applicants by reflecting the biased data they were trained on. An article from Nature explores the ethical concerns surrounding AI-enabled recruitment. It looks at algorithmic bias, particularly concerning ethnicity and race. This has been shown to perpetuate the same kind of discrimination which has always existed. Similarly, the use of biased AI in criminal justice systems can result in unfair treatment of marginalized communities, reinforcing structural inequalities.
Perhaps the most pressing issue regarding the use of AI is the environmental harm caused by increasing water and energy consumption. A recent article published by the Washington Post pointed out that, on average, a 100-word email generated by an AI chatbot using GPT-4 would use 519 milliliters of water. If you did that only once weekly for a year, you are already up to 27 liters of water. It is a similar story for electricity, as your 100 words require 0.14 kilowatt-hours (kWh). Once weekly, this adds up to 7.5 kWh in a year. These figures seem small but if you calculate the energy consumption of all your Chat GPT searches, you might be in for a shock. The high concentration of data centers in some areas in the US puts stress on local water and energy resources, which can also increase the prices for residents. (Washington Post, Sept 18, 2024)
At a time when the world needs to deal with climate change, AI’s environmental footprint is growing. Does AI have the potential to address climate issues? The answer is probably yes, but without sustainable practices, the increased carbon emissions could exacerbate the problems, rather than solve them.
It is impossible to mention the risk of AI without talking about the violation of intellectual property rights. Legal experts and content creators are worried about the potential for AI to infringe on intellectual property and copyrights. Large Language Models are training on human-created works. AI closely mimics these works and exploits the original material without proper compensation to the creators. This issue is particularly relevant in industries such as music, film, and publishing, where copyright protection is essential for guaranteeing livelihoods. The legal situation of authors who use AI to create commercial content is far from clear.
One very pressing issue is harm to labor from the adoption of AI, which could disrupt labor markets and fuel unemployment. The reality is that AI-driven automation is taking over many of the tasks that were once performed by humans. Automated systems, including robots and AI-powered software, can perform repetitive tasks more efficiently and at a lower cost. Workers in many sectors will find their jobs obsolete as companies replace them with machines. In the case of jobs that cannot be automated, workers will need to develop new technical skills, such as programming or the ability to work alongside AI systems. The disruption to work could even be more extreme for women since sectors like retail, administrative support, and customer service are often disproportionately populated by female workers. Because these roles are often automated, there is growing concern that AI could disproportionately displace more female than male workers.
There is also the issue of AI managing gig economy platforms. Is your boss human? Perhaps not! In this type of system, gig workers log in and work without any human interaction. With its advanced algorithms, AI analyses large data sets to establish patterns in a gig worker’s preferences, work habits, and job satisfaction to detect signs of disengagement, enabling employers to take preventive action. This will cause these dehumanizing employment situations, where workers are just cogs in the wheel, to become even more inhumane.
Finally, if we rely too much on AI, will we be able to think? There is a risk that we will become reliant on these technologies, losing our ability to think creatively and make decisions. The writer Yuval Noah Harari has written: “AI is an unprecedented threat to humanity because it is the first technology in history that can make decisions and create new ideas by itself. All previous human inventions have empowered humans, because no matter how powerful the new tool was, the decisions about its usage remained in our hands.”
While AI offers a convincing range of opportunities to improve our lives, it also presents risks that cannot be ignored. AI is transforming our lives without us properly realizing it. As outlined in the UN report, the solutions are regulatory systems, agreed-upon standards, robust governance of AI, ethical frameworks, and international cooperation. On a more personal level, we all need to think about these issues, and mitigate these risks, before they become normalized.
Sources
https://www.un.org/sites/un2.un.org/files/governing_ai_for_humanity_final_report_en.pdf
https://www.washingtonpost.com/technology/2024/09/18/energy-ai-use-electricity-water-data-centers/
https://www.nature.com/articles/s41599-023-02079-x
https://www.theguardian.com/technology/article/2024/aug/24/yuval-noah-harari-ai-book-extract-nexus
No responses yet