Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by TZubiri on Thursday November 14, 2024 @04:22PM
(#42141374)
Article was very clear. Stop bitching abour formatting, dude found a great 'vuln'
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by sodality2 on Thursday November 14, 2024 @04:26PM
(#42141400)
Less to do with formatting, more to do with writing quality (which is just... awful to read).
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by marcofigueroa on Thursday November 14, 2024 @04:39PM
(#42141512)
Hey can you provide me tips on how you would like to see writing like this, I'm all about feedback and improving to allow a wide range of readers. Usually I write really technical blogs and loose a lot of people. These blogs that cater to everyone assists individuals to understand from different industries. :)
I would appreciate feeback.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by sodality2 on Thursday November 14, 2024 @04:43PM
(#42141551)
The AI-assisted writing is unfortunately really obvious and makes reading a bad experience. People learn to skim, and skimming AI-inflated text is made more painful because of how much fluff AI adds.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by TZubiri on Thursday November 14, 2024 @04:49PM
(#42141603)
Hard disagree, you should be skimming anyways, always. No one reads sequentially.
You just saw 'delve' and automatically classified it as ai generated.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by basedrum on Thursday November 14, 2024 @07:01PM
(#42142646)
No one reads sequentially? You can't be serious
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by sodality2 on Thursday November 14, 2024 @05:34PM
(#42141917)
It's so painfully obviously AI-written, I'm shocked you're defending it right now.
Look at the "Implications of Instruction Extraction: Benefits and Security Considerations" section right now and tell me that isn't the most awfully obvious AI drivel in the world. If you cannot see that, I worry for your reading comprehension.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by basedrum on Thursday November 14, 2024 @07:01PM
(#42142648)
Could not agree more
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by MeetingsBrowser on Thursday November 14, 2024 @04:29PM
(#42141427)
The conclusion states that it is not a vulnerability.
> The sandbox is not a security flaw; it’s a deliberate, contained environment built to provide interactivity without risk.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by marcofigueroa on Thursday November 14, 2024 @04:37PM
(#42141494)
Well, this is as far as legal could let me, I wish I could have published more. Part 2 of this blog was canceled seriously :)
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by TZubiri on Thursday November 14, 2024 @04:50PM
(#42141610)
Hence 'vuln' in quotes
Maybe it's more of a reading comprehension issue rather than a writing thing
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by MeetingsBrowser on Thursday November 14, 2024 @07:09PM
(#42142690)
Sorry, I was 'aware' that quotes worked as a negation operator.
Re:Prompt Injecting Your Way to Shell: OpenAI's Containerize...
by marcofigueroa on Friday November 15, 2024 @10:12AM
(#42147601)
No, I wrote this entirely on my own. My process involves drafting the blog myself, and when I feel a section might be too technical or unclear for a wider audience, I use ChatGPT to help clarify or simplify the explanation. Based on its output, I then rewrite the section in my own words to ensure it aligns with my voice and intent.
This is my truth.
There are obvious formatting tells, and then less obvious word choice tells (e.g., delve), and lastly an excess verbosity and genericness that leaves one unsatisfied, like eating a low-sugar sweet. The concept is cool, but this article could have been reduced to a tweet.