Privacy and Performance: Google Bard vs ChatGPT

In the realm of large language models (LLMs), two popular options are Google Bard and ChatGPT. While both are capable of generating high-quality text, there are notable differences between them, both technically and in terms of privacy. This article provides an overview of the technical distinctions between the two and delves into the legal implications associated with each tool, shedding light on their respective privacy practices.

Technical Similarities

From a technical standpoint, both Google Bard and ChatGPT serve similar purposes. They excel in generating rough drafts, working well with structured data and curated prompts. They are valuable for creative projects and providing reliable answers to questions based on historically validated information.

However, it is crucial to note that neither tool should be used without human oversight in decisions that have significant legal or consumer-related consequences. They are not suitable for producing final published drafts of journalism or reliable news media. Additionally, their reliability in deciphering unstructured or untested information should not be assumed.

Data Sources

Google Bard has been trained on an “infiniset” of data selected to enhance its conversational abilities and has real-time internet access. This allows Google Bard to retrieve more accurate and relevant information. On the other hand, ChatGPT has been trained on an undisclosed dataset until 2021 and does not currently offer internet access to all users. However, ChatGPT Plus users can enable a beta feature called “Web browsing” in the Settings, which grants limited internet access using the ChatGPT-4 Model.

While the prospect of granting complete internet access to chatbots may seem enticing, previous experiences with internet-connected chatbots have shown limitations. It is safer to assume that both systems will face similar challenges. The internet, as a source of training data, is not inherently safe, and algorithms still struggle to effectively discern safe and unsafe content.

Privacy Considerations

In terms of privacy, neither Google Bard nor ChatGPT provides an ideal solution. Both systems have access to highly sensitive information and limited control over user data. However, when comparing their privacy practices, ChatGPT may be considered relatively more privacy-protecting. Let’s explore the specifics:

Google Bard’s Privacy Practices

Google Bard claims to be more privacy-protecting than ChatGPT, stating that it does not collect or store personal information. It assures users that their browsing activities are not tracked, and their data is not used for advertising purposes. However, these claims are misleading, as Google Bard requires users to be logged into a Google account, providing identifiable information. Conversations with Google Bard are also sent to human reviewers for annotation, and although account information is removed, the content of conversations may still be retained and used for various purposes across Google’s ecosystem.

ChatGPT’s Privacy Practices

While ChatGPT and Google’s Bard both seem intrusive, they both actually are. Data breaches at ChatGPT have already occurred, as have careless actions by business players who exposed the system to confidential company data or source code. Simply put, it is what has been reported. Remember that once anything is entered, it cannot be removed, and there is a significant probability that the system will utilise your confidential information for another reason if someone asks a question that is similar to or might use your material as an answer.

OpenAI, however, has demonstrated that it at least publicly promises to perform better thanks to its policy of working in public. The fact that ChatGPT has an API makes it accessible to more plugins and has been implemented into more programmes, which is another advantage. Notably, ChatGPT is extensively used in the low- and no-code communities, making it simpler to prototype systems with and accessible to a larger audience. Perhaps this is what Bard meant but did not fully clarify when it suggested ChatGPT is more accessible.

Choosing the Right Tool

The choice between Google Bard and ChatGPT depends on specific requirements and preferences. If real-time internet results are essential, Google Bard may be a suitable option. However, it lacks an API and production capabilities. On the other hand, ChatGPT offers accessibility and can be used immediately in production, making it more convenient for developers and no-code users.

It is worth noting that numerous alternatives are emerging, focusing on user privacy and providing platforms for future development. Privacy-focused tools like Privacy AI’s Private GPT offer enhanced privacy protection by monitoring and filtering confidential data before it reaches ChatGPT.

While the development of large language models is promising, privacy concerns must be taken seriously. Striking a balance between innovation and privacy protection is crucial as we navigate this evolving landscape.

Conclusion

Google Bard and ChatGPT are both powerful language models with their respective strengths and limitations. By understanding their technical differences and privacy practices, users can make informed decisions based on their specific needs. Privacy-conscious alternatives and tools are emerging, ensuring a safer and more privacy-protecting future for large language models. As the field continues to evolve, it is vital for professionals to be aware of the impact of these technologies and prioritize privacy considerations.

Source