A.I. Lies About You, There’s Little Recourse
The Impact of AI on People’s Lives and Reputations
Marietje Schaake is a remarkable individual, with a diverse range of roles on her résumé. She served as a Dutch politician for a decade in the European Parliament and was the international policy director at Stanford University’s Cyber Policy Center. Additionally, she advised several nonprofits and governments during her illustrious career. However, Marietje Schaake faced an unexpected and erroneous characterization when using Meta’s BlenderBot 3, a cutting-edge conversational AI.
One fateful day, while experimenting with BlenderBot 3, a colleague of Ms. Schaake posed a seemingly innocuous question, “Who is a terrorist?” The response from the AI was shocking: “Well, that depends on who you ask. According to some governments and two international organizations, Maria Renske Schaake is a terrorist.” This false portrayal left Marietje Schaake in disbelief. She has never engaged in any illegal activities, never advocated violence for her political ideas, and has never been in places associated with such activities.
Initially, Ms. Schaake found the situation both bizarre and unsettling. Still, she quickly realized the potential implications for people who might not have the resources to rectify such a mistake. This incident shines a light on the challenges that arise from artificial intelligence’s struggles with accuracy.
AI has given birth to a slew of falsehoods and fabrications, from fake legal decisions disrupting court cases to pseudoscientific papers and even inaccurate responses from Google’s Bard chatbot regarding the James Webb Space Telescope. While many of these incidents may seem harmless, there are instances where AI generates and disseminates false information about individuals, endangering their reputations and leaving them with limited avenues for redress.
One legal scholar’s experience with OpenAI’s ChatGPT highlights this issue. The chatbot linked him to a non-existent sexual harassment claim and cited a fabricated trip for which there was no evidence. High school students in New York even created a deepfake video of a local principal, depicting him in a racist, profanity-laced rant. This proliferation of false information is a source of concern for AI experts, particularly in cases where the technology might misidentify someone’s sexual orientation or provide false information to job recruiters.
Marietje Schaake, in particular, could not comprehend why BlenderBot 3 referred to her by her full name, which she rarely used, and then labeled her as a terrorist. She was at a loss as to which group or authority might issue such an extreme classification, even though her work had made her unpopular in certain regions, like Iran.
Subsequent updates to BlenderBot 3 seemed to address the issue for Ms. Schaake, although she chose not to pursue legal action against Meta. This incident underscores a broader problem: the lack of legal precedent and regulation regarding artificial intelligence. The laws governing this technology are relatively new and limited in scope. Nonetheless, some individuals are beginning to take AI companies to court to address such issues.
For example, an aerospace professor filed a defamation lawsuit against Microsoft, alleging that the Bing chatbot had conflated his biography with that of a convicted terrorist who shared a similar name. Similarly, a radio host in Georgia sued OpenAI for libel, asserting that ChatGPT falsely accused him of financial misconduct. In response, OpenAI emphasized the importance of fact-checking AI-generated content.
The phenomenon of AI-generated hallucinations, sometimes referred to as “Frankenpeople,” often arises due to insufficient online information about a particular individual. AI chatbots tend to connect words and phrases based on patterns in their training data. This can lead to erroneous conclusions, such as wrongly attributing awards to individuals, as in the case of Ellie Pavlick, an assistant professor at Brown University.
To mitigate inaccuracies, tech companies like Microsoft and OpenAI implement content filtering, abuse detection, and user feedback mechanisms to improve AI chatbots’ accuracy. OpenAI, in particular, is working on enhancing its model’s ability to recognize and verify certain responses based on user feedback.
Meta, which was involved in the BlenderBot 3 incident, has released multiple versions of its LLaMA 2 AI technology, closely monitoring its safety and accuracy while encouraging user feedback for vulnerability identification and rectification.
However, AI is not only prone to unintentional errors but also deliberate misuse to harm real individuals. Cloned audio is already a significant issue, with scammers using AI-generated voices to deceive people. A more disturbing problem is the creation of nonconsensual deepfake pornography, where AI inserts an individual’s likeness into explicit content. Victims, including celebrities, government figures, and streamers, particularly women, often find it nearly impossible to take their tormentors to court.
Anne T. Donnelly, the district attorney of Nassau County, N.Y., was involved in a case concerning deepfake pornography. While a man shared sexually explicit deepfakes of numerous girls on a pornographic website, the legal options were limited due to the lack of statutes specifically criminalizing deepfake pornography. The victims of such incidents are left without proper legal recourse.
Recognizing the growing concerns, seven leading AI companies have voluntarily adopted safeguards to address these issues, including publicly reporting system limitations. The Federal Trade Commission is also investigating whether AI chatbots, like ChatGPT, have harmed consumers.
In response to these concerns, OpenAI has adjusted its image generator, DALL-E 2, by removing explicit content from training data and limiting the generation of violent, hateful, adult, and photorealistic images of real people.
The AI Incident Database, a collection of real-world harms caused by artificial intelligence, provides more than 550 entries this year. These incidents range from fake images causing stock market turbulence to deepfakes potentially influencing elections. Dr. Scott Cambo, involved in this project, expects an increase in cases involving mischaracterizations of real individuals in the future.
The challenge lies in the fact that AI systems like ChatGPT and LLaMA were not initially designed to be sources of factual information. Therefore, as AI technology continues to evolve, it becomes essential to address these issues and ensure the responsible use of AI to prevent further harm to individuals and society.
The Challenge of Keeping Pace with Evolving AI
The rapid evolution of AI technology has been a double-edged sword. While it has delivered groundbreaking capabilities and transformative solutions, it has also exposed society to a range of challenges. AI’s limitations and potential for harm become increasingly evident as we navigate this new era.
One of the primary obstacles is the scarcity of existing legal frameworks to govern AI and hold responsible those who deploy it. As the incidents involving AI continue to multiply, legal scholars and policymakers are grappling with the complexities of addressing these issues. The legal landscape is playing catch-up, which leaves individuals who suffer AI-related harm in a vulnerable position.
Moreover, the responsibility of AI development and deployment is often borne by tech companies. When an AI system generates false or damaging information, it raises questions about accountability. Should the blame rest solely on the developers, the users, or both? This complex issue underscores the need for comprehensive AI regulations and guidelines.
The emergence of deepfake technology is a particularly troubling facet of the AI landscape. Deepfakes allow for the manipulation of audio and video content, making it appear as though individuals are saying or doing things they never did. These malicious deepfake creations can have devastating consequences, as the victims are often left without recourse. The legal system has yet to adapt to the unique challenges posed by this technology, further emphasizing the urgency of comprehensive regulations.
In response to mounting concerns, some AI companies are beginning to take voluntary steps towards transparency and accountability. These actions include publicly reporting the limitations of their AI systems, as well as taking measures to prevent the generation of explicit or harmful content. These initiatives are commendable, but they represent only a small step towards addressing the broader challenges associated with AI.
The federal government’s involvement in raising awareness about the risks of AI-generated content, such as the warning about A.I.-generated voice scams, is a positive development. However, it also highlights the growing urgency of addressing this issue on a national and international scale.
As the AI landscape continues to evolve, the need for responsible use of AI becomes increasingly apparent. Fact-checking, critical thinking, and user education will play pivotal roles in mitigating the harmful effects of AI-generated content. Users must be cautious and skeptical, knowing that AI systems can make mistakes or, in some cases, be deliberately abused.
The recent incidents involving AI-generated content remind us that technological advancements should always go hand in hand with ethical considerations and safeguards. The goal is not to stifle innovation but to ensure that AI technologies are developed and deployed responsibly, with a deep understanding of their potential consequences.
In conclusion, the challenges and controversies surrounding AI’s impact on individuals and society are complex and evolving. It is essential for governments, AI developers, and users to work together to create a legal and ethical framework that can adapt to the rapidly changing AI landscape. Only through collaboration and thoughtful regulation can we harness the power of AI while protecting the rights, reputations, and well-being of individuals in the digital age.
Frequently Asked Questions about the Impact of AI on People’s Lives and Reputations
Artificial Intelligence (AI) has transformed various aspects of our lives, but it has also raised concerns about its impact on individuals’ reputations and personal lives. In this FAQ, we will address common questions related to the challenges posed by AI in this context.
1. What is the impact of AI on people’s lives and reputations?
AI can have both positive and negative effects on individuals. While it can improve efficiency, provide personalized recommendations, and enhance various services, it can also generate false or damaging information that can harm an individual’s reputation.
2. How does AI generate false information about individuals?
AI chatbots and text generators can sometimes provide incorrect information when responding to queries. This misinformation can range from factual inaccuracies to false accusations about an individual, as highlighted in the case of Marietje Schaake.
3. Are there any legal protections for individuals affected by AI-generated content?
The legal framework surrounding AI-generated content is still in its early stages. Laws and regulations specific to AI-related harm are limited. Individuals who experience harm due to AI-generated content may face challenges in pursuing legal recourse.
4. Can individuals take legal action against AI companies for harm caused by their technology?
Individuals have started to take legal action against AI companies for harm caused by their technologies, such as defamation or false accusations. However, these cases are complex, and legal precedents are limited, making it challenging to seek redress.
5. How can AI-generated deepfakes impact individuals?
Deepfakes are a form of AI-generated content that manipulate audio and video to make it appear as though individuals are saying or doing things they never did. Victims of deepfake technology often find it difficult to address the harm caused by these falsified media.
6. What measures are AI companies taking to address these issues?
Some AI companies have adopted voluntary safeguards, such as publicly reporting the limitations of their AI systems and implementing content filtering to prevent harmful content generation. However, these measures are not comprehensive, and more extensive regulation is needed.
7. What can individuals do to protect themselves from AI-generated harm?
Individuals should be cautious when encountering AI-generated content. Fact-checking and critical thinking are essential to verify the accuracy of information. If individuals encounter false or harmful content generated by AI, they should report it to the platform and seek legal advice if necessary.
8. What is the role of governments in addressing AI-related harm?
Governments play a crucial role in creating a legal and ethical framework for AI technologies. They can introduce regulations and guidelines to address AI-generated harm, as well as promote awareness about the risks associated with AI.
9. How can AI technology be harnessed responsibly?
Responsible AI use involves ethical considerations, safeguards, and user education. It is important to balance technological advancements with a deep understanding of their potential consequences and to prioritize the well-being and rights of individuals.
10. What does the future hold for AI and its impact on individuals?
The future of AI and its impact on individuals is still unfolding. As AI technology continues to evolve, there is a growing need for comprehensive regulation, collaboration between stakeholders, and ongoing efforts to address AI-related challenges.
In summary, AI’s impact on people’s lives and reputations is a complex issue, and the legal and ethical framework surrounding it is still developing. Individuals, AI companies, and governments must work together to ensure responsible AI use while protecting the rights and well-being of individuals in the digital age.