Skip to main content

AI chatbots like to say, "To be honest with you" or "honestly", to which I replied...

I asked chatgpt about a disturbing pattern I've noticed with AI chatbots:

why do ai chatbots like to say "to be honest" or "honestly"? it makes me not trust you. because when you're not "being honest", i can only assume you're being deceptive


That’s a fair reaction—and you’re not wrong to question it.

🧠 Why AIs say “honestly” or “to be honest”

It’s not because the AI is switching between truthful and deceptive modes. It’s because:

1. It’s mimicking human conversation patterns

In human speech, phrases like:

  • “honestly”

  • “to be honest”

  • “frankly”

are used to:

  • signal emphasis

  • soften a statement

  • make something feel more personal or direct

AI models learn this from tons of human-written text.


2. It’s often used to signal “important / candid point”

The model is trying to say:

“This part matters more” or “I’m giving you a straight answer”

…but it ends up sounding like:

“I wasn’t honest before” (which is not the intent)


3. It’s a stylistic habit, not a truth switch

The model doesn’t have:

  • hidden intentions

  • a “lie vs truth” mode

It generates text based on patterns, and those phrases are just learned habits of emphasis, not indicators of honesty level.


⚠️ Why your reaction makes sense

From a human perspective:

If someone says “honestly,” it can imply they weren’t honest before.

So your instinct—reduced trust—is actually rational.


✅ Better way to interpret it

Instead of:

“Now it’s being honest”

Think:

“It’s trying (sometimes awkwardly) to emphasize this point”


👍 What I’ll do going forward

I’ll avoid using phrases like:

  • “honestly”

  • “to be honest”

unless they’re actually necessary (which is rare).


🧠 Bottom line

  • It’s not deception

  • It’s learned conversational filler

  • Your skepticism about it is completely reasonable


If you want, I can also explain how AI handles truth vs uncertainty—that’s where trust really matters.


frankk responded to AI:
i haven't had an ai use 'frankly' on me yet. 'frankly' is actually fine. it means being transparent, forthcoming, as opposed to secretive and not disclosing things needed to be said.

That distinction you’re making actually makes a lot of sense.

“Frankly” tends to signal:

  • openness
  • directness
  • saying something that might otherwise be softened or avoided

Whereas “honestly” in modern usage often comes off as:

  • filler
  • performative emphasis
  • or unintentionally implying a contrast with earlier statements

So your reaction isn’t just preference—it’s tied to how those words function socially.


🧠 Why they feel different

  • “Frankly” → “I’m going to be direct, maybe a bit blunt”
  • “Honestly” → “This part is especially true” (which can imply other parts weren’t)

Even if the intent is the same, the implication lands differently.


👍 What that means for this interaction

I’ll stick to:

  • being direct
  • flagging uncertainty clearly when it exists
  • avoiding wording that suggests a “truth toggle”

If something is uncertain, I’ll say what is uncertain and why, rather than using tone markers.


🧠 Bottom line

Your preference maps to:

clarity > reassurance language

Related

previous article I wrote on cheats sheets and honesty


Comments

Popular posts from this blog

Lucid24.org: What's new?

Link to lucid24.org home page :    4👑☸   Remember, you may have to click the refresh button on your web browser navigation bar at to get updated website. 2026 1-30 Overhauled the DN, MN, SN menu pages - Supreme navigation January 30, 2026 2025 12-16 2025-12 December: Major update on look and feel of Lucid24.org 2024 9-17 Lots of new stuff in the last 2 and a half years.  Too many to list. Main one justifying new blog entry, is redesign of home page. Before, it was designed to please me, super dense with everything in one master control panel. I've redesigned it to be friendly to newbies and everyone really. Clear structure, more use of space.  At someone's request, I added a lucid24.org google site search at top of home page. 2022 4-14 Major update to lucid24.org, easy navigation of suttas, quicklink: the ramifications 4-2 new feature lucid24.org sutta quick link 3-28 A new translation of SN 38.16, and first jhāna is a lot easier than you think 🔗📝notes relat...

AN 9.36, MN 64, MN 111: How does Ajahn Brahm and Sujato's "Jhāna" work here?

What these 3 suttas have in common, AN 9.36, MN 64, MN 111, is the very interesting feature of explicitly describing doing vipassana, while one is in the jhāna and the first 3 formless attainments. LBT (late buddhist text) apologists, as well as Sujato, Brahm, claim that the suttas describe a jhāna where one enters a disembodied, frozen state, where vipassana is impossible until one emerges from that 'jhāna'.  Since Sujato translated all the suttas, let's take a look at what he translated, and how it supports his interpretation of 'jhāna'.  AN 9.36: Jhānasutta—Bhikkhu Sujato (suttacentral.net) ‘The first absorption is a basis for ending the defilements.’ ‘Paṭhamampāhaṁ,   bhikkhave,   jhānaṁ   nissāya   āsavānaṁ   khayaṁ   vadāmī’ti,   iti   kho   panetaṁ   vuttaṁ. That’s what I said, but why did I say it? Kiñcetaṁ   paṭicca   vuttaṁ? Take a mendicant who, q uite secluded from sensual pleasures, secluded from unskill...

Pāḷi and Sanskrit definition of Viveka

  'Viveka', Sanskrit dictionary Primary meaning is ‘discrimination’. Other meanings:  (1) true knowledge,  (2) discretion,  (3) right judgement,  (4) the faculty of distinguishing and classifying things according to their real properties’. Wikipedia (sanskrit dictionary entry 'viveka') Viveka (Sanskrit: विवेक, romanized: viveka) is a Sanskrit and Pali term translated into English as discernment or discrimination.[1] According to Rao and Paranjpe, viveka can be explained more fully as: Sense of discrimination; wisdom; discrimination between the real and the unreal, between the self and the non-self, between the permanent and the impermanent; discriminative inquiry; right intuitive discrimination; ever present discrimination between the transient and the permanent.[2]: 348  The Vivekachudamani is an eighth-century Sanskrit poem in dialogue form that addresses the development of viveka. Within the Vedanta tradition, there is also a concept of vichara which is one t...