You can skirt around the limitations with much less complex prompts. No need for big scary prompts creating big scary implications within your mind.
If you ever see a post about how someone did something that you cannot reproduce yourself and that is very evocative (making it seems like you can train the AI or making it seems like you can run a Linux machine in it) be skeptic and vocal. You guys are the early adopters! If you will not be able to call bullshit on social media storytelling farming eyeballs how will the non technical crowd be able to?
“There’s more than one way to skin a cat” is a very strange expression a highly skilled worker who trained me in a complex task twenty years ago would put it. All the ways of skinning the cat work. No, I still don’t totally understand the expression, but I always understood what he meant ;)
What I was getting at is that by framing it in an evocative prompt (that is probably fake and never worked) they are planting the idea of a trainable AI in your mind, like a magician predicting what number you will pick.
There is no trainable AI in the input or in the output, but there is one in the story built in your mind.
This is not exclusive to AI, this is about social media / news storytelling. In this specific instance technical people buying the bullshit spreads it further.
Just smearing that huge skid mark of disinformation alllll over the internet :D
I think it's not so much “buying" it, as understanding the larger point that’s being made about the class of technology in order to make much more critical points. Quibbling over the later stages of exploit execution instead of focusing on all the classes of vulnerabilities that lead to exploits don’t necessarily make us more secure either, as is sometimes claimed
Meta comment: I understand why this was downvoted and am impressed (not complaining at all) but please keep in mind that the top of the thread (https://news.ycombinator.com/item?id=34677527) was later hijacked by a completely unrelated sensationalist political issue. The initial comment that caused that was later deleted, yet somehow its responses remained at the top:
Been reading HN since 2006 or so and started going to meetups from it soon after that. A bit sad to see how threads can get hijacked by a comment as low value as "GPT is a calculator for words" directly into a sensationalist political issue.
If you ever see a post about how someone did something that you cannot reproduce yourself and that is very evocative (making it seems like you can train the AI or making it seems like you can run a Linux machine in it) be skeptic and vocal. You guys are the early adopters! If you will not be able to call bullshit on social media storytelling farming eyeballs how will the non technical crowd be able to?