The Balance to be Struck
Earlier this year, a law firm based in California sued Open AI for using millions of people’s personal data without permission to develop the core of ChatGPT and other artificial intelligence (“AI”) platforms.(1) The issue of protecting people’s privacy has been hotly debated for many years due to technological advancements. However, with the advent of numerous AI products in recent years, privacy issues have become even more prevalent and urgent. One of the biggest concerns arises from a fundamental goal of AI, namely the urge to improve the quality of AI platforms by allowing them to be exposed to large amounts of data without impinging on personal information. Achieving this goal is surely easier said than done—enabling AI products to amass online data would almost inevitably result in the leakage of private data. Nevertheless, this begs the question of whether our priority should lie in enhancing the quality of AI products or protecting people’s personal data.
On one hand, in order for AI platforms to flourish and unleash their full potential, they need to have full access to high-quality data. An article published by the MIT Sloan School of Management stated that “focusing on high-quality data...would unlock the value of AI for sectors, such as health care, government technology, and manufacturing.”(2) As a matter of fact, according to a publication by the European Commission, a police agency was able to reduce rates of burglary by 33% and violent crimes by 21% by enabling their AI system to access large amounts of Open Data in relation to crime occurrences and frequencies.(3) Furthermore, according to a 2023 academic journal published by the Society for Industrial and Applied Mathematics, the “role of data in building AI systems has recently been significantly magnified by the emerging concept of data-centric AI”.(4) Rather than basing AI systems on models, which focuses on structuring algorithms with limited amounts of data, creating AI systems entrenched in big data and Open Data has resulted in more efficient and productive AI platforms that provide users with more accurate and faster responses. Hence, AI platforms empowered by high-quality data ultimately benefit society by improving the overall quality of life.
On the other hand, protecting people’s privacy is tantamount to increasing the quality of AI models. The World Economic Forum (“WEF”) stated in a recently published article that “data privacy is essential to [gaining] the trust of the public in technological advances”. (5) In addition, the WEF also stated that “some top-rated online services and products could not work without personal data to train their AI algorithms.” However, without an AI platform that uses private information in a “secure and non-invasive” manner, the public would be highly reluctant to use AI platforms, which would eventually lead to the extinction of AI models. In fact, a survey conducted by the European Consumer Organization in 2020 showed that “45 to 60% of Europeans agreed that AI will lead to more abuse of personal data.” This result demonstrates that the lack of trust in AI could prove to be fatal to the existence of AI in the short run. Therefore, it is crucial for the future of AI that society creates a “privacy-respecting” AI model that "safeguards both the autonomy and privacy of users.”
Undoubtedly, AI models depend on data amount and quality to deliver salient results. Consequently, their existence will hinge on integrating privacy protection into their design. In the end, a balance needs to be struck between obtaining data and protecting sensitive information. Because AI plays a prominent role in almost all aspects of our current society, it is inevitable that online services and products will continue to obtain large datasets that include personal information to teach and improve AI algorithms. However, we will need to establish safeguards to “improve the acquisition, management, and use of data” to ensure that the systems are only using “data with clear and informed user consent.” (6) Ultimately, our end goal would be to maximize the potential of AI models in a manner that does not infringe on people’s fundamental right to privacy.
Bibliography
1 Freedman, R. (2023, June 29). Lawsuit seeks to curb AI rush, hold OpenAI accountable for stolen data. Legal Dive. https://www.legaldive.com/news/OpenAI-class-action-lawsuit-internet- scraping-privacy-violations-chatgpt- law/654332/#:~:text=A%20California%20law%20firm%20on,that%20are%20quickly%20bein g%20incorporated
2 Brown, S. (2022, June 7). Why it’s time for “data-centric Artificial Intelligence.” MIT Sloan. https://mitsloan.mit.edu/ideas-made-to-matter/why-its-time-data-centric-artificial-intelligence
3 AI and Open Data: A crucial combination. (2018, July 4). https://data.europa.eu/en/publications/datastories/ai-and-open-data-crucial-combination
4 Data-centric AI: Perspectives and challenges - siam publications library. (n.d.-b). https://epubs.siam.org/doi/abs/10.1137/1.9781611977653.ch106
5 Artificial Intelligence Design must prioritize data privacy. World Economic Forum. (n.d.). https://www.weforum.org/agenda/2022/03/designing-artificial-intelligence-for-privacy/
6 Artificial Intelligence Design must prioritize data privacy. World Economic Forum. (n.d.). https://www.weforum.org/agenda/2022/03/designing-artificial-intelligence-for-privacy/