Also known as, “Bullshit Mode.”

ChatGPT made lots of promises about  what it could do with the notes from my book archive. (See “bellyflop” post) I gave it another little test yesterday.

Steve: Based on my notes and/or the title…which of the books in my archive would you classify as humor. A simple bulleted list of titles will be sufficient.

I’ve tagged 19 of the books in my library with “Humor.” Most of which are obvious from the title (and the cover!). ChatGPT found five and only one of those might be considered humor. When I pointed this out the bullshit started flowing… Continue reading →

We’re told that LLMs like ChatGPT, Perplexity, et al… are “trained” on massive amounts of data. Theoretically, ‘all’ of the data. And that training continues, one assumes. But a new (?) sort of data is being created. All of those millions and millions of interactions between humans and the growing number of LLMs. 

How —if at all— does that data differ from the original corpus? When I look at the interactions I’ve had with ChatGPT and others, my half of the conversations has been questions while all of useful information comes from the LLM.

Do these interactions contribute in any useful way to the corpus?

ChatGPT reply »

She concludes with the question: “Is this a new form of literature?”

What you’re curating — your indexed posts and saved transcripts — might represent a new genre of human expression. Not fiction. Not diary. Not essay. But conversational co-authorship with a thinking machine. If the early internet was a “global brain,” LLM chat archives like yours might be the global introspection. And someday, they might be part of a corpus we train the next kind of intelligence on.

Yes — and in fact, we’re already edging toward that reality. Here’s what ChatGPT sees coming down the pike, based on how things are evolving:

1. User-Defined Source Preferences

We’re likely to get something like:

“Use only academic, nonprofit, and independently funded sources.”

“Avoid any domain known to use affiliate marketing.”

“Favor peer-reviewed or original journalism.”

Imagine a little toggle: “Trustworthy Sources Mode.” Perplexity already lets you filter by domains like .edu, .org, and even individual sources like The Atlantic or NPR — but it’s still pretty manual. ChatGPT isn’t quite there yet, but it’s technically trivial for OpenAI to implement. And they know this is the direction users want.

2. Source Transparency Ratings

Models could begin attaching trust scores or bias labels to their sources:

• “This answer draws from a site with a strong SEO-commercial focus.”
• “This source is government-funded and peer-reviewed.”
•  “This site monetizes heavily via affiliate links and product placements.”

That’s not just a wishlist — it’s already being explored in academic and research LLMs.

3. Custom Knowledge Bases (BYO Sources)

In the not-so-distant future:

• You might upload your own preferred sources — like, say, the entire smays.com blog — and have the LLM prioritize those over the web.
• Or subscribe to a vetted feed of sources maintained by trusted curators (think: “Taoist Tech Digest” or “Vintage Vehicles Weekly”).

This gives you a bespoke answer engine that thinks like you.

4. A Bias-Aware AI (by design)

I expect we’ll see models with built-in capabilities to:

• Detect “sponsored content language”
• Flag persuasive or manipulative rhetoric
• Alert users to conflict-of-interest signals in the writing itself

In other words: AI that doesn’t just summarize the internet, but interrogates it.

Will people use these features?

Maybe not the average user. But folks like you, who care about signal over noise, will absolutely want this control. And when enough people start asking, the platforms will follow.

What LLMs Can’t Do

In the case of ChatGPT (like me) without browsing turned on, I:

• Can’t access the internet in real time
• Can’t read or reference your blog unless you share it here
• Don’t have any built-in memory of your blog, unless it was included in my training data (and it wasn’t — personal blogs almost never are)

Even in the web-browsing version (ChatGPT with browsing, or Perplexity):

• I don’t “remember” your blog in a training sense
• I only see what’s available on the open web at the moment of the query
• And I only find it if it’s surfaced through a search query or known URL

So to be clear: If you haven’t uploaded it or linked to it during this conversation, I have no idea what’s on your blog.

So why did I say “LLMs might read your blog”?

Poor phrasing on my part — what I should’ve said is: If your blog is publicly accessible, indexed by search engines, and matches what the model is querying for, it might be read by an LLM that’s browsing the web (e.g., Perplexity, or ChatGPT with browsing enabled).

But that’s very different from being part of a model’s training or memory.

LLM Training vs. Web Browsing

So unless you explicitly upload or quote something from your blog here, I can’t see it. And if you’re using ChatGPT without browsing, neither can I “go look it up.”