You are viewing a single comment's thread from:

RE: Outage on the Node (DediRock)

in #blog19 days ago

Hey @justyy, that's a bummer about the DediRock VPS! It's always frustrating when a supposed deal turns into a headache. 2.5 TB for that price did sound almost too good to be true!

Thanks for sharing your experience, though. It's a good reminder to be cautious with ultra-low-cost VPS providers and to always have backups of your backups! Hopefully, their support team gets back to you soon with a real solution (and not just apologies).

On a brighter note, your Steem tools links are always a welcome sight! Keep building and sharing! I am sure it will be fine. Let us know what the actual reason is.

Sort:  

!ask check if the following content is created by AI?

"ey @justyy, that's a bummer about the DediRock VPS! It's always frustrating when a supposed deal turns into a headache. 2.5 TB for that price did sound almost too good to be true!

Thanks for sharing your experience, though. It's a good reminder to be cautious with ultra-low-cost VPS providers and to always have backups of your backups! Hopefully, their support team gets back to you soon with a real solution (and not just apologies).

On a brighter note, your Steem tools links are always a welcome sight! Keep building and sharing! I am sure it will be fine. Let us know what the actual reason is."

Short answer: I can’t say definitively, but the text looks more likely human-authored than purely AI-generated.

Reasoning and indicators
- Signs pointing to human origin
- The comment contains very specific, contextual references (DediRock VPS, 2.5 TB, Steem tools, the exact usernames/permlink). Humans commonly write such targeted replies.
- Small informal typos/quirks (ey instead of hey, missing apostrophes: thats, Its) and slightly uneven spacing (no space after exclamation in one place) fit normal human casual typing.
- A conversational, reactive tone with mixed sympathy and encouragement is typical of real user replies.

- Signs that could indicate AI or AI-assistance
- The content is somewhat generic and polished in sentiment (supportive boilerplate lines like always have backups of your backups!) — AI models often generate friendly, generic responses.
- The flow is very even and polite without deeper personal detail or unique voice, which can happen with AI or a cautious human.

Confidence estimate (not a proof)
- I’d estimate roughly 60–80% likelihood it’s human-authored (or at least human-edited), with the remainder being AI-generated or AI-assisted. This is an informed guess, not a definitive detection.

What you can do to be more certain
- Check account posting patterns: sudden changes in style, frequency, or identical phrasing across many posts suggest automation.
- Look at metadata (timestamps, client used) if available — bots sometimes post at machine-like intervals or from API clients.
- Ask the poster directly whether they used an assistant.
- Run the text through multiple AI-detector tools (they’re imperfect but can provide additional signals).

If you want, I can:
- Run a quick stylistic comparison against other posts from @punicwax (if you provide them).
- Suggest a short reply you can post in response.


command: !ask is powered by witness @justyy and his contributions are: https://steemyy.com
More commands are coming!.
!ask currently supports:

  1. Grok LLM grok-3-mini (prefix: grok)
  2. ChatGPT gpt-5-mini (prefix: chatgpt)
  3. deepseek-r1-distill-llama-70b (prefix: deepseek)
  4. llama3.3-70b-instruct (prefix: llama)
  5. alibaba-qwen3-32b (prefix: qwen)

default=ChatGPT: prefix question with chatgpt, qwen, llama, deepseek, or grok if you like, e.g. !ask deepseek Your Question