Is Google’s Bard the ChatGPT killer? For privacy’s sake, I sure hope not

Riley Hughes
3 min readMay 16, 2023

--

I tried Midjourney to create the cover photo for this post “robot peering in through a window” and… yeah, apologies for the nightmares.

I’ve seen a dozen posts about Bard vs ChatGPT. Everyone seems to be breaking them down in terms of their features. But before using Bard, I had a simple question: Would my Bard convos be used against me? Will my conversations be the raw material for producing ad-targets on my back?

According to the privacy policies of Bard and ChatGPT, and here’s what I learned.

To set the stage: data protection and privacy really matter. No matter how many “not medical advice” or other warning labels these softwares give, people will continue using them for deeply personal matters. If these sensitive queries create a target on my back, I’ll be more cautious about what questions I ask.

In addition, many businesses disallow use of ChatGPT or similar tools on work laptops for fear of models being trained on proprietary business data. (Which, by the way, doesn’t make sense, given these same businesses allow employees to use Google, which is also training its own AI models on what you search.)

Privacy & data protection aren’t just ethical, but have real implications to product adoption. So how does Bart compare to ChatGPT across a few important dimensions?

Bard vs ChatGPT

First of all, everything you submit to Bard is “read, annotated, and processed” by “trained reviewers” and is stored for “up to three years”. This is in contrast to OpenAI — they “retain new conversations for 30 days and review them only when needed to monitor for abuse, before permanently deleting”. ChatGPT is the clear winner here.

In addition to manual review, your submissions to Bard are also used in automated fashion to “improve and develop Google products and services and machine-learning technologies”. To be fair, the same thing is done in ChatGPT, unless you pay them for the business subscription where they claim “end users’ data won’t be used to train our models by default”. Github CoPilot, which is powered by OpenAI, has a business subscription with similar guarantees.

This brings us to the reason for my exploration: will my Bard conversations be used against me, to target me for advertising? According to Google, “Bard conversations are not being used for advertising purposes, and we will clearly communicate any changes to this approach in the future.”

But as a company with $225 billion in advertising revenue last year (at least, that’s what Bard told me, so I’m not sure how confident I should be in that answer!). The stakes are too high — it’s obvious that they will monetize Bard in the future. Fortunately, you can feel reasonably confident your Bart usage in the near-term won’t be used for advertising, because when they retain your conversations for up to 3 years, they disconnect them from your account. The conversations associated with your Google account are stored for 18 months by default, but you can configure that in settings, and delete specific conversations, similar to your Youtube history, etc.

Ultimately I feel like this is another chapter in the saga of ad-supported, free products vs subscriptions. I pay for a Pro ChatGPT license, and will continue doing so and using ChatGPT as my primary companion in my day-to-day.

--

--