The last couple of months have been full of news related to large language models. For example, OpenAI announced a data breach which allowed some users to see titles from the other active user’s chat history. On the other hand, Samsung employees were exposed to confidential meeting minutes and source code due to prompts sent to ChatGPT. OpenAI, then, has reiterated its commitment to protect user’s privacy and security. Such incidents have raised questions in people’s minds as well, which is ultimately why the Future of Life Institute sent out an open letter to the AI community, warning of the potential risks that powerful AI systems bring. In response to the open letter, AI veterans such as Andrew Ng and Yann LeCun asserted their disagreement with the calls of halting the research on AI systems.
In response to the OpenAI data breach, Garante, the Italian data protection authority, has intervened and the Office of the Privacy Commissioner of Canada has launched an investigation as well. With an increasing number of powerful AI systems being developed and tested, there are certain implications which need to be considered, especially in regards to societal implications, regulation, governance, etc. Goldman Sachs predicted that generative AI systems could add up to 7% to the GDP of the world in ten years, although it will also lead to the loss of 300 million jobs.
OpenAI has highlighted that ChatGPT users should avoid sending sensitive information in their prompts in order to avoid future data breaches from occurring. Similarly, GPT-4 is reported to have impressive performance data and theAI industry is currently working on developing a multi-modal prompt combined with images and texts. The massive amounts of data collected for the development of these systems raise questions regarding consent, copyright issues, data accuracy, and more. The UK government’s whitepaper recently outlined the interventionist approach towards creating AI systems as well.
In conclusion, the development of AI technology and its powerful applications bring many ethical implications as well. While these new AI systems provide many benefits to our world, it is important to remember to have proper governance and regulation in place to protect user’s data and privacy. OpenAI and others in the industry are already championing these efforts, but users need to be adequately educated to recognize potential risks and take the necessary steps to protect their data.
