How your ChatGPT caricature is feeding the AI machine | ABC News In-depth

Key Concepts

• AI-Generated Caricatures: Images created by AI chatbots (specifically ChatGPT) based on user data.
• Data Collection & Monetization: The practice of AI companies gathering user data and potentially profiting from it.
• Data Set Tiering: The concept of varying data quality and access based on subscription levels.
• Data Privacy Concerns: Risks associated with sharing personal information with AI platforms, including potential resharing.

The Rise of AI-Generated Self-Portraits & Initial Concerns

A recent trend involves users generating caricatures of themselves using AI chatbots like ChatGPT. These images are created based on a combination of the user’s chat history with the platform and any photos they provide. If insufficient data exists, the AI prompts the user for further personal details about their life. While seemingly harmless, AI experts express significant concern regarding this practice, viewing it as a strategic method for AI companies to acquire more user data. The core issue is that this data fuels the AI’s performance and encourages continued user engagement.

Data as Currency: Collection and Potential Resale

The fundamental principle driving these AI chatbots is data. The more information they possess, the more effectively they function, and the more appealing they become to users. Once personal information is submitted to these platforms, it becomes part of a proprietary data set owned by the AI company. A key risk highlighted is the potential for this data to be shared with other users, potentially without the original user’s consent or knowledge.

OpenAI’s Stance & the Reality of Data Access

OpenAI publicly states it does not sell personal data. However, Professor Dustwan points out a critical nuance: data can be effectively shared through tiered subscription models. He explains that “the quality of the information provided is quite uh different at these tiers. So if you wanted to find out information about an individual, it is just know how much are you willing to pay.” This implies that access to more detailed or comprehensive data sets is granted to paying subscribers, effectively creating a market for information derived from user data. The cost of accessing information about an individual is directly proportional to the subscription level.

Expert Advice: Caution and Data Minimization

Professor Dustwan advises cautious engagement with AI chatbots. While acknowledging their utility, he stresses the importance of being mindful of the personal information shared. The concern stems from the uncertainty surrounding the ultimate destination and usage of this data. As the speaker notes, after initially being curious to generate their own caricature, they became more cautious after consulting with experts.

Strategic Data Gathering

The initial statement frames the caricature generation as “probably one… very strategic method of collecting more data that can be sold later on.” This highlights the deliberate nature of the data collection process, suggesting it isn’t merely a byproduct of AI functionality but a core component of the business model.

Conclusion

The trend of AI-generated self-portraits, while appearing playful, represents a significant data privacy concern. AI companies are actively incentivizing users to provide more personal information, which is then incorporated into proprietary data sets. While direct sales of data may be prohibited, tiered subscription models effectively allow for data access to be monetized. Users should exercise caution and minimize the personal information shared with these platforms to mitigate potential risks.