**Key Differences Between Google Bard and ChatGPT:**
When comparing Google Bard and ChatGPT, there are some notable functional differences between the two language models. However, they are not significantly different from a technical standpoint. Both systems claim to generate human-quality text, and while Google Bard promotes itself as more accurate and relevant, ChatGPT emphasizes its accessibility.
**Functionality of Both Systems**
Both Google Bard and ChatGPT are effective for generating rough drafts and working with structured data and curated prompts. They are also suitable for creative projects that don’t have legal implications. Additionally, these tools can provide reliable answers to objective facts with historical validation. However, it is crucial to note that they should not be used without human oversight in decisions that could have legal consequences or significant impacts on consumers. They are not recommended for producing final drafts of journalism or reliable news media. Furthermore, they cannot be considered reliable when it comes to interpreting unstructured data or untested information.
Google Bard has been trained on an extensive dataset selected by Google to enhance its dialogue capabilities. It also has real-time access to the internet, allowing it to retrieve accurate and relevant information. On the other hand, ChatGPT has been trained on an undisclosed dataset up to 2021 and does not provide internet access to all users. However, ChatGPT Plus users can access the web browsing feature by enabling it in the settings.
While the idea of granting these language models unrestricted internet access seems appealing, it is important to acknowledge the limitations. Previous internet-connected chatbots have faced challenges in discerning safe and appropriate content for training data. Both Google Bard and ChatGPT may struggle with this issue. Assuming that all internet data is safe for training purposes is problematic, and it is concerning that these systems cannot effectively filter out harmful content such as hate speech and pedophilia. Implementing safeguards is necessary for the successful development of these services.
Neither Google Bard nor ChatGPT excel in terms of privacy protection. Both systems have access to sensitive information without sufficient user control. However, ChatGPT can be considered more privacy-protective, as explained below.
**Privacy Practices of Google Bard**
Google Bard claims to prioritize privacy by stating that it does not collect or store personal information about its users. It assures users that it doesn’t track their browsing activities or utilize their data for advertising purposes. However, these claims can be misleading. To use Google Bard, users must be logged into a Google account, making it clear that Google has knowledge of their identities. Although users can delete their prompts, conversations are not eliminated entirely.
According to Google’s policies, conversations are sent to human reviewers for annotation, and Google retains these annotated conversations. Even when the activity tracker is turned off, Google retains the information for 48 hours for feedback processing. Thus, any personally identifiable information, confidential or proprietary details, or sensitive conversations will be stored by Google.
The claim that the saved text is labeled as “anonymous” because user information is removed from the conversation is a form of privacy greenwashing. It fails to meet legal privacy requirements and does not provide adequate protection. Google’s history of data retention and usage raises doubts about their commitment to deleting data.
Furthermore, by using Google Bard, users agree not only to the Generative AI Additional Terms of Service but also to the company-wide Terms of Service. This implies that the data retained by Google can be used across the company.
Although neither Google Bard nor ChatGPT offers satisfactory privacy protection, ChatGPT can be considered relatively more privacy-protective than Google Bard.