Some people are now so dependent on chatbots they want to transfer info and chat histories from one bot to another.

[highplainsdem]

There were news stories today about Google's Gemini now having widgets to encourage users of other bots to switch their addictions to Gemini:
https://techcrunch.com/2026/03/26/you-can-now-transfer-your-chats-and-personal-information-from-other-chatbots-directly-into-gemini/

When it comes to AI chatbots, there’s currently a war on for consumer attention. All the big chatbot providers are looking to increase their user count and, in a minor coup for itself, Google just made it significantly easier for users of those other chatbots to defect to Gemini.

-snip-

The memory feature works like this: Gemini will suggest a prompt that the user can enter into their current chatbot, which will then generate a response that can be copied and pasted back into Gemini. In this fashion, Gemini coaches the user on what kinds of information it would be helpful to know about them, while also helping facilitate the transmission of that information back into its own archive.

“Once you import these memories, Gemini will understand the same key facts you’ve shared with other apps, like your interests, your sibling’s name, or where you grew up,” the company says. “Instead of starting over from scratch, you can quickly get Gemini up to speed on what matters most to you.”

When it comes to importing chat histories, Google says that all you need is to upload them in a zip file. It’s relatively easy to export chat logs via zips from most chatbots — including from ChatGPT and Claude. This allows users to “seamlessly pick up right where you left off,” the company says. Google says users also have the ability to search through those old chats.

Of course the AI bros and the venture capitalists backing them want people to give all their information to chatbots, and to feel they can't get by without them. Learned helplessness via AI. This sort of transfer of data helps Google's data gathering so much, and shows them through all those chat histories how their chatbot can keep the new users engaged and manipulate them. One article about Google that I saw today mentioned Anthropic having already done something similar, and presumably the other AI companies will follow. With addicted users providing more and more training data, maybe spending their days moving back and forth between different bots that know everything about them, flies caught in an AI web.

[LearnedHand]

1. Well, yes, but there's also more to it

People in many industries use chatbots for routine work, such as coding, proofreading, and even as a thought partner to explore concepts. I can see clear value in moving chat history from one chat tool to another. I’m not saying this is the best way to work, just that people DO work that way. And it’s often at the insistence of the employer. My former employer leaned heavily on the tech and professionalis staffs to use the internal chat ai tools.

Totally agree with you about the commercial tools and the monster companies behind their development.

[Hugin]

3. How many people are actually doing this?

The article implies that it’s almost everyone.

I don’t know anyone who is. Not even one. I even see many fewer publicly using assistants like siri and Alexa.

[LearnedHand]

4. The research organization where I worked used it heavily

As did researchers from across the industry, as reported in technical conferences and trade journals. This was serious work, not confessing secret thoughts to an ai therapist. I’m just saying it’s not all just people addicted to chatbots. I myself found it to be not useful for my work but I saw very impressive demos of people using the tool the right way.

[highplainsdem]

6. I don't know what type of research you mean, but if those researchers are using AI to replace usual

search and to provide sources and summarize what's found at those sources, the results are unfortunately likely to be riddled with errors unless the researchers painstakingly checked every detail of the AI's results. And that use of AI, when not caught by peer review, has meant publication of some of those error-filled research results.

The hallucination problem and need to check every detail of AI results will pretty much cancel out time saved with AI, which is probably the top selling point for AI use.

I'm not saying people using AI for research aren't sincerely convinced those AI tools are helping. Or that the results can't look impressive, if not examined too closely.

But what I've read about AI and productivity suggest that it doesn't increase productivity as much as users like to believe. And their believing it helps so much, and becoming reluctant to give up using it, does suggest work-related AI addiction for at least some users:
https://www.democraticunderground.com/100221071042

[highplainsdem]

7. Thanks, Hugin! That's good news!

[synni]

2. I've been using Gemini recently, and it's been a godsend

It has pointed me toward the types of doctors that I need to see in the near future, while my current doctors just shrug, say they don't know, and leave me to suffer.

Gemini also prevented me from getting food poisoning, because I was going to handle a cooking task incorrectly.

Gemini is just like the internet. It has power, so you need to learn how to use the power in a constructive way. Just like the internet, you have to keep your skepticism and wits about you as you consult it.

For someone like me who is visually impaired, it has been a wonderful thing to be able to do research quickly and have the results read to me aloud I can tell when it's making mistakes, but I have found a lot more mistakes in the questionable "knowledge" that people have been uploading to the web for decades.

[highplainsdem]

5. Please be extremely careful about trusting chatbots for medical advice. They're not truly intelligent,

they hallucinate, and they're sycophantic because they're designed to keep users engaged if not completely addicted.

[pecosbob]

8. Also keep in mind people's 'private' chats are now showing up repackaged as content on Youtube.

[highplainsdem]

9. Yes. And I've never forgotten reading about an early incident with ChatGPT where users were shown

content from other users' chats.