Artificial Intelligence (AI) has moved beyond the realm of science fiction and into our daily lives, bringing both benefits and risks. While discussions about the potential for AI to pose existential threats to humanity have captured headlines, the true concern lies in the immediate and tangible harm AI technologies are already causing. From wrongful arrests to algorithmic discrimination, the focus needs to shift from sensationalism to grounded science that investigates real-world dangers.
Unmasking the real threats
Rather than succumbing to the allure of sensationalist rhetoric, the focus should be on the real harms that AI technologies are inflicting today. Wrongful arrests, expanding surveillance networks, and the proliferation of deep-fake pornography are not hypothetical scenarios; they are actual consequences of AI tools currently available on the market. The rise of discriminatory practices in housing, criminal justice, and healthcare, as well as the spread of hate speech and misinformation, highlights the pressing issues tied to AI’s present impact.
Dismantling the hype
Many AI companies tend to prioritize dramatic narratives over tangible harms when discussing AI’s potential risks. The recent statement from the Center for AI Safety, co-signed by industry leaders, regarding the risk of AI-driven extinction, has sparked concerns about misplaced priorities. The narrative surrounding “existential risk” deflects attention from the urgent need to address the very real harms already emerging.
The need for clear definitions
The term “AI” itself has become an ambiguous catch-all, encompassing various meanings ranging from a subfield of computer science to an enigmatic magical solution for businesses. This ambiguity complicates meaningful discussions. Today, text synthesis models like ChatGPT dominate the AI landscape. These models generate coherent text but lack comprehension and reasoning abilities, making them more akin to sophisticated chatbots than genuine intelligent entities.
The problem with synthetic text
While text synthesis models produce convincing text, they lack true understanding, making their outputs potentially misleading and harmful. The proliferation of synthetic text exacerbates misinformation and bias present in their training data, amplifying societal prejudices. The challenge lies in distinguishing synthetic text from credible information sources, a task that becomes more difficult as synthetic text spreads.
AI as a solution? Or a detriment?
Text synthesis technology is often hailed as a solution to various societal gaps, such as education, healthcare, and legal services. However, the reality is different. The technology often exploits content created by artists and authors without proper compensation, and the process of labeling data for AI training involves underpaid gig workers subjected to harsh working conditions. Furthermore, the pursuit of automation results in job losses and precarious employment, especially in industries like entertainment.
The role of Science-driven policy
AI policy must be rooted in solid research and genuine concern for the welfare of society. Unfortunately, many AI publications come from corporate or industry-funded sources, leading to questionable research practices. It is imperative to shift the focus to research that investigates the actual harms AI systems perpetuate. These include the unchecked accumulation of data and computing power, environmental costs of AI training, and exacerbation of social inequalities.
Refocusing on genuine research
Policy-makers must prioritize rigorous research that delves into the harms and risks of AI, along with the consequences of delegating authority to automated systems. This research should encompass social science analysis and theory-building, fostering a deeper understanding of the societal impacts of AI. Policies based on such research will ensure that the focus remains on addressing the real-world issues that marginalized communities face due to AI technologies.
As AI continues to shape our world, the emphasis should shift from sensationalized existential threats to the immediate harm that AI technologies are causing. Wrongful arrests, algorithmic discrimination, and the spread of hate speech are among the serious consequences of AI tools today. Focusing on grounded science and solid research will enable policymakers to address these pressing issues and ensure that AI technologies benefit society without causing undue harm.
