Google’s Bard Is My Counselor Now

 

Siri has egg on her face

Anthony Lawrence (Pcunix)

TECHNOLOGY

DALLE Image by author Robed Wizard Pointing Down a Road

Google’s Bard AI came out of invite only mode on May 18th and I immediately fired it up to try it out. Since then, I usually turn to it for information rather than normal Googling.

So far, it has only let me down once. Other than that, I prefer it to normal Googling.

Bard itself can tell you when you should use it.

I asked, “When should I use Bard instead of normal Googling?” Bard replied:

You should use Bard instead of normal Googling when you need:

• More comprehensive answers: Bard can access a wider range of information than Google Search, including information from books, articles, and websites. This means that Bard can often provide more comprehensive answers to your questions.
• More creative output: Bard can generate different creative text formats, like poems, code, scripts, musical pieces, email, letters, etc. This means that Bard can be a valuable tool for writers, artists, and musicians.
• A conversation partner: Bard can hold a conversation with you, even if your questions are open ended, challenging, or strange. This means that Bard can be a helpful tool for brainstorming, problem solving, and learning.

I then asked, “When should I use normal Google instead of Bard?”

• You should use normal Google instead of Bard when you need:
• Quick answers: Google Search is often faster than Bard for simple questions.
• Specific information: Google Search can be more specific than Bard. For example, if you are looking for information on a specific topic, Google Search can help you find articles, websites, and books that are specifically about that topic.
• Fact-checking: Google Search can be helpful for fact-checking information that Bard provides.
  For example, if Bard provides you with an answer to a question, you can use Google Search to find other sources that corroborate that answer.

Here are some examples of when you might want to use Google Search instead of Bard:

• You need to find a quick answer to a simple question.
• You are looking for specific information on a topic.
• You want to fact-check information that Bard provides.

Hmm. Bard itself tells me that it’s not always the best choice. Maybe that’s because I ask weird questions?

“Do I need ChatGPT-4 now that I have Bard?”

Whether you need ChatGPT-4 now that you have Bard depends on your specific needs.

Bard and ChatGPT-4 are both large language models (LLMs) that can generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way. However, there are some key differences between the two models:

• Bard is trained on a dataset of text and code that is updated regularly, while ChatGPT-4 is trained on a dataset that was last updated in 2021. This means that Bard has access to more up-to-date information.
• Bard is designed to be more comprehensive and informative in its answers, while ChatGPT-4 is designed to be more creative and engaging. This means that Bard may be a better choice for tasks such as research and writing, while ChatGPT-4 may be a better choice for tasks such as storytelling and generating marketing copy.
• Bard is free to use, while ChatGPT-4 is a paid service.

Yes, Bard is free and ad-free right now, but how long will that last? I expect that Google Bard will be replacing Google Search for a lot of people. Google ad revenue has dropped recently, so Bard Ads would not surprise me.

Of course, there is no free. We are constantly reminded that if you are offered anything for free, you are being mined for information. What Google gleans from my Bard queries may be used to improve Bard itself, but it may also be used to target me for advertising. Hey, let’s ask Bard what it thinks about that!

“What does Google do with the data it mines from my questions here?”

Google uses the data it mines from your questions here to improve the quality of its products and services. For example, Google may use your questions to:

• Train its machine learning models to better understand natural language.
• Identify trends in user behavior so that Google can develop new products and services that meet the needs of its users.
• Personalize your experience with Google products and services.
• Improve the accuracy of search results.
• Prevent spam and abuse.

I Then asked ChatGPT-4 about its privacy protections.

1. OpenAl does not use the data sent via the ChatGPT API to improve thr models. This means your individual conversations are not being used train the models further.
2. OpenAl may collect and store personal data as provided by users, but only for the purpose of providing and improving services. This could include troubleshooting, maintaining the safety and security of the services, and conducting data analysis and research.
3. OpenAl does not sell personal data collected from users. The use and disclosure of such data is generally limited to purposes directly relate to providing and improving the services.
4. OpenAl implements data protection strategies such as anonymizatior and encryption to safeguard data.

However, the specifics may vary depending on OpenAl’s current policies, and those might have been updated after my last training cut-off in September 2021.

I’m not sure what to make of that. It doesn’t use my questions but it does use my personal data?

Time to ask Bard another question.

“Can you identify trends in user behavior if I am using IOS Private Relay and have tracking turned off?”

No, I cannot identify trends in user behavior if you are using lOS Private Relay and have tracking turned off. IOS Private Relay is a feature that allows you to encrypt your internet traffic and route it through two different servers, which makes it more difficult for websites and other online services to track your activity. When tracking is turned off, I am not able to collect any data about your browsing habits, which means that I cannot identify trends in your behavior.

IOS Private Relay is a valuable privacy feature that can help to protect your online activity from being tracked. If you are concerned about your privacy, I recommend that you enable OS Private Relay and turn off tracking.

Thanks for recommending that I protect myself from you, Bard, but gosh I wish I had something Bard-like from Apple rather than from you. So, I asked “Could Apple make Siri better without compromising user privacy?”

Yes, Apple could make Siri better without compromising user privacy. Here are a few ways Apple could do this:

1. Improve Siri’s accuracy and reliability. Apple could improve Siri’s accuracy and reliability by training it on a larger and more diverse dataset of text and code. This would help Siri to better understand natural language and to generate more accurate and reliable responses.
2. Make Siri more personalized. Apple could make Siri more personalized by allowing users to train it on their own data. This would help Siri to better understand the user’s preferences and to generate more personalized responses.
3. Give users more control over Siri’s privacy settings. Apple could give users more control over Siri’s privacy settings by allowing them to choose what data is collected and how it is used. This would help users to feel more comfortable using Siri and to protect their privacy.

Encrypt Siri’s data. Apple could encrypt Siri’s data to protect it from unauthorized access. This would help to ensure that user data is not shared with third parties without the user’s consent.

1. Be transparent about how Siri works. Apple could be more transparent about how Siri works by publishing more information about its algorithms and data collection practices. This would help users to understand how their data is being used and to make informed decisions about whether or not to use Siri.

By following these steps, Apple could make Siri better without compromising user privacy.

I asked ChatGPT-4 the same question and got a much more detailed answer.

On-Device Processing: By processing data locally on the user’s device instead of sending it to a server, user data can be kept more private. Apple has been making strides in this direction by using the processing power of its devices to enable Siri to handle more tasks locally.

Differential Privacy: This is a technique that adds noise to datasets, so that any data that is sent back to Apple for analysis doesn’t tie back to individual users.

Federated Learning: This is a machine learning approach where the model is trained across multiple devices while keeping data localized. Essentially, instead of sending data to the cloud, a global model is sent to the devices, and learns from data on the device. Then, only the model updates (and not the data) are sent back and aggregated to improve the global model.

Data Minimization: This involves collecting only the data that is strictly necessary for functionality. The less data that’s collected, the less there is to potentially be compromised.

Opt-in Data Collection: Giving users the choice to contribute data for improving services. This respects user privacy by not assuming consent.

Transparent Policies: Clearly communicating to users what data is being collected and for what purpose it is being used. Transparency builds trust and allows users to make informed decisions regarding their privacy.

Regular Security Updates: Keeping the system and services secure through regular updates to protect against new vulnerabilities and threats.

Customized Processing: Apple could improve Siri’s performance by allowing more customization and user feedback. By learning from corrections and preferences directly stated by the user, Siri could become more useful without needing to analyze sensitive data.

By employing a combination of these approaches and technologies, Apple can continue to enhance Siri’s capabilities while maintaining a strong commitment to user privacy.

So why don’t they? Bard told me it might be a matter of resources. Huh? If anybody has the resources, it should be Apple. Siri has been embarrassing for some time now, but Bard, ChatGPT and the rest push that to an extreme level now.

C’mon Apple: Step up your game!

You might think that having several thousand followers here means big money. Actually, it usually gets me only a dollar or so a day. That is why donations, one time or monthly, are always appreciated. See https://ko-fi.com/pcunix