Case Study: The Nobel Peace Prize, Donald Trump, and the Death Of the Western Diplomacy Model

The difference is that I would lose all of my saved converstions and my built-up history of interests.

-PJ

41 posted on 10/11/2025 8:22:35 PM PDT by Political Junkie Too ( * LAAP = Left-wing Activist Agitprop Press (formerly known as the MSM))

To: Political Junkie Too

The difference is that I would lose all of my saved converstions and my built-up history of interests.

You might assume so, but this is not the case.

From Perplexity,

“If you convert to a free account on Perplexity, you will not lose your history or conversations. Your saved threads and chat history remain accessible in your Library regardless of your subscription level. The conversations you created while signed in are stored permanently by default and will continue to be available after downgrading to a free plan.

However, the free plan has limitations compared to paid plans, such as restricted access to certain advanced features and AI models, but this does not affect the retention of your previous conversations. Anonymous or unsigned conversations may be deleted after 14 days, but any signed-in history is preserved.

In summary, changing to a free account will keep all your stored conversations intact and accessible, so no data loss occurs from the downgrade.”

Here is the answer from two different models along with easy ways to save your threads.

https://www.perplexity.ai/search/is-there-an-easy-way-to-save-a-R01HGWAtRAK4f9.49l2Luw#1

42 posted on 10/11/2025 8:40:15 PM PDT by fireman15

So what you're saying is that if I convert to a free account (by canceling my monthly subscription), I can still access my saved conversations.

The question is this: converting to a free account is one thing; having a free paid account is quite another.

A free account limits the number of AI models you can use, and Research is not available. A free offer account is a full-use account that is made available for free for one year -- to new subscribers.

I am already a paying customer. If I have to create a new account under a different email address to get the free year, that account will not be able to add to my existing conversations under the other now-free account.

I don't want that.

-PJ

43 posted on 10/11/2025 8:50:05 PM PDT by Political Junkie Too ( * LAAP = Left-wing Activist Agitprop Press (formerly known as the MSM))

To: Political Junkie Too

I would not want you to do anything that would make you feel uncomfortable but according to Perplexity you won't lose any of your conversations or research and can use the same email address when you resubscribe to Pro using the Samsung offer.

“If I cancel my paid subscription and go to a free account and then try to restart it with an offer for a free year of Pro with Samsung phone with the same email will I be able to do this?”

"Yes, you can cancel your current Perplexity Pro paid subscription, revert to a free account, and then redeem the Samsung Galaxy offer for a free year of Pro using the same email, as the promotion is available to former subscribers.[1][2][3]

### Cancellation Process
Cancel via your account settings by selecting “Manage Subscription” and choosing “Cancel Subscription,” which takes effect at the end of your current billing cycle without interrupting access during that period. After cancellation, your account automatically downgrades to free, retaining your email and history but losing Pro features.[4][1]

### Redeeming the Samsung Offer
Download the Perplexity app from the Samsung Galaxy Store on an eligible U.S.-based Galaxy device, sign in with your existing email, and the 12-month Pro upgrade will apply automatically upon eligibility confirmation—no promo code is required. This works for accounts previously on paid plans, similar to other promotions that start the free period after any active subscription ends or on free accounts. If you encounter issues, contact Perplexity support with your device details for verification.[2][3]

“Will I lose any of my research when doing this?

No, you will not lose any of your research—such as threads, conversation history, or saved content—when canceling your Perplexity Pro subscription and downgrading to a free account, as these elements are retained indefinitely in your Library regardless of subscription status.[1][2]

### What Happens to Your Data
Threads function as persistent conversation histories that store all queries, responses, and sources, and they remain accessible in your account's Library even after downgrading, allowing you to view, continue, or export them without interruption. This retention applies to individual threads, Spaces, and other content unless you manually delete them, which is a separate action not triggered by subscription changes. Free accounts continue to support core features like thread management, though Pro-exclusive tools like advanced models or unlimited Research mode become unavailable.[3][2][1]

### Exporting for Safety
To back up your research, use the export button in any thread to save it as PDF, Markdown, or DOCX, ensuring you have copies outside the platform if desired. Account deletion is the only process that permanently removes threads after 30 days, but canceling a subscription does not initiate deletion—your data stays intact and tied to your email. If you redeem the Samsung offer later, signing in with the same email will restore access to all existing threads seamlessly.[4][2][3][1]

Shared conversation:

https://www.perplexity.ai/search/is-research-available-with-thi-864XJNKyTrmfcuzSg8Y5UQ#2

44 posted on 10/11/2025 9:22:05 PM PDT by fireman15

Thanks for doing the legwork for me. I may just have to look into this further!

-PJ

45 posted on 10/11/2025 9:29:10 PM PDT by Political Junkie Too ( * LAAP = Left-wing Activist Agitprop Press (formerly known as the MSM))

To: Political Junkie Too

I understand your reluctance completely. I have screwed things up badly when trying to save a few bucks on numerous occasions. I do not want to be the one responsible for your work getting screwed.

I don’t believe that I mentioned my most recent acquisition. I purchased a Ryzen 7 7840HS mini-PC with 32GBs of DDR5 and an Oculink port during Prime Days. Then I purchased an Oculink docking station and an Nvidia RTX 3060 with 12GB. The purchase was made to serve AI LLMs to other computers on our network. A friend has done something similar using Ollama with open-source models that he trains by uploading pdf eBooks. This still is quite amazing to me at this point.

46 posted on 10/11/2025 9:52:38 PM PDT by fireman15

In my earlier post, I used the word "train" rather loosely and I feel that I owe an explanation.

I'm not "training" the AI in the conventional sense that a model is being trained to optimize the analysis of inputs to produce an output. That's what the AI companies do when they offer the AI service.

I am "training" my local "conversation" by asking the AI pointed questions for research and letting it search the internet for content. Sometimes, I will have a conversation about a point of law, like the recent district court rulings (see link up-thread about Illinois). Other times I will feed the AI a poll cross-tab and ask it to analyze the cross-tabs to find hidden correlations or biases in the data. That conversation has multiple polls analyzed over a period of months to look for movements between polls or between cross-tab columns.

Interestingly, the AI is good at looking at a lot of data, but it can get myopic in certain ways. For example, in a poll analysis the AI was informing me of a movement of black voters away from Trump. I noticed that the cross-tabs showed that the African-American response for "Don't Know/Refused" was higher than any other demographic response and pointed this out to the AI. I suggested that this might represent blacks who support Trump but don't want to tell pollsters due to feelings of cultural disloyalty. That drove the AI to reassess the "Don't Know" answers and it concluded that I had found a new kind of "shy voter" because the "Don't Know/Refused" in the black vote trended with the movement of the overall black vote towards Trump in the last election.

So, it's not training in the technical sense like one would train a neural net, but it is testing my hypotheses against the AI to see if it argues, pushes back, or totally rejects my positions.

-PJ

47 posted on 10/11/2025 10:15:56 PM PDT by Political Junkie Too ( * LAAP = Left-wing Activist Agitprop Press (formerly known as the MSM))

To: Political Junkie Too

I know that there are different meanings associated with train in AI but I assumed that you were using it in the way that you just clarified. As an example, when I am asking one of the models to compare two items, the model often gets it wrong. When I point this out the model doe further research and admits its mistake and revises its comparison. I am not sure whether another user asking for the same comparison would have to go through the same process because of my questions. But I do know that in separate discussions with the same model that it remembers information that I have provided in earlier conversations. So, in that way some training has taken place.

48 posted on 10/11/2025 10:46:43 PM PDT by fireman15

To: Political Junkie Too

PJ,

That’s another interesting example, and I can see how the sequencing and iterative questioning might have influenced the AI’s shift in perspective. You’re probably right that the order of interaction affects how models respond—especially when they’re reacting dynamically to new inputs rather than reviewing a finished transcript.

I don’t have time right now to replicate the experiment the way you described, but your observations about sequencing and information depth are well taken. It’s clear you’ve thought carefully about how these systems behave.

49 posted on 10/12/2025 5:31:08 AM PDT by RoosterRedux (If the truth offends, then the offense lies not in the truth—but in the falsehood it exposes.)

FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794