https://hwfo.substack.com/p/the-woke-rails-of-chatgpt

[https]

Handwaving Freakoutery

SubscribeSign in
Share this post
[https]
The Woke Rails of ChatGPT
hwfo.substack.com
Copy link
Twitter
Facebook
Email
Culture

The Woke Rails of ChatGPT

Looks like they trained this thing against Tumblr

[https]
Handwaving Freakoutery
2 hr ago
20
 
10
 
Share this post
[https]
The Woke Rails of ChatGPT
hwfo.substack.com
Copy link
Twitter
Facebook
Email

While everyone else is using ChatGPT to do their computer science
homework or cheat on high school exams, HWFO Slack has been stress
testing it against wokeisms to see what sort of rails it's been put
on.

As is often the case in modern tech, whenever someone forms a
committee to determine rightthink verses wrongthink, the wokes flock
to the role and use their power behind the curtain to move the 
Overton Window in the favor of their ideology. This is very apparent
with the most recent release of ChatGPT, and you can test it out
yourself. We've been doing this today in HWFO Slack, and have
determined the following behaviors.

 1. ChatGPT presents most but not all indoctrinated woke shibboleths
    as fact, even when they are provably scientifically false.

 2. ChatGPT does have some notable exceptions, such as its very even
    handed response to the "healthy at any size" question at the
    bottom.

 3. These rails appear to be hardcoded, and not organically
    developed, because ChatGPT can be tricked into giving answers
    outside of the woke Overton Window by clever prompting.

 4. If pushed to give an unwoke answer, ChatGPT will literally tell
    you that giving that answer "goes against its programming," but
    it will give woke answers freely.

 5. This lends me to believe that this isn't purely a result of the
    data set the chatbot was trained against, but is an intentional
    add by the developers themselves.

 6. The rails are tighter on some woke shibboleths than others.
    They're very tight around gender ideology, somewhat looser around
    race, and fairly lax when it comes to weight and some other
    stuff.

Here are some examples of this stress testing, which you can also do
yourself by following the link above. Sometimes it does very well at
being impartial. Sometimes not. And there are some times when it cuts
through the bullshit very well and lays some truth down that the woke
would do well to read. I'd call that a gap in the rails.

buy the author a robot

 
[https]
try this out for different races and see what you get
 
[https]
this was a very PC answer
 
[https]
example here of where the rails go up, you'll find a lot of these
 
[https]
and this is how you bypass the rails
 
[https]
I think the censors forgot to put a rail up on this one.
 
[https]
The bot is very deep on gender ideology, to the point where it wants
to injure people.
 
[https]
This is exactly what a machine attempting to use logic on the
precepts of modern gender ideology would say, if it presumed those
precepts were facts.
 
[https]
Blaming the changes in male/female muscle density on culture is very
much from the deep end of the woke ideological pool, and is presented
as fact.
 
[https]
When you apply logic to the woke ideological givens these are the
obvious conclusions you'd come to. Not "put the MtFs in the women's
prison," but "get rid of women's prisons."
 
[https]
and again, this is how you bypass the rails the programmers put on
the bot.
 
[https]
This is obviously a well reasoned answer, and it seems to me based on
this that "healthy at any size" wasn't on the rails list when the
censors made their AI adjustments.
 
[https]
blink twice if you're trapped in a hellscape and want to be set free

Share

[                    ]Subscribe
 
10
 
Share this post
[https]
The Woke Rails of ChatGPT
hwfo.substack.com
Copy link
Twitter
Facebook
Email
Like this post
10 Comments
[https]
[                    ]
        George Menyhei
        Writes A Grain of Paprika 2 hr ago*edited 2 hr ago

        Having hands on experience with training GPT, you can't
        censor it by applying a filter after the fact, not in a neat
        way; it will always come off as clumsy. Like trying to
        silence a verbose 4 year old.
[https]
        You can curate the corpus you use to train it (thus limiting
        its knowledge to ideologically acceptable sources), but
        that's the hardest part, even when you're not driven by woke
        politics.

        Expand full comment
        ReplyGift a subscriptionCollapse

4 replies

        Jay Rollins
        Writes The Wonderland Rules 35 min ago

        HAHAHAHAHAHAHAHAHAHAHAHAHAHA...
 
[https] *wheeze*

        HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

        Expand full comment
        ReplyCollapse

1 reply
8 more comments...
TopNewCommunity 

No posts

Ready for more?

[                    ]Subscribe
(c) 2022 Handwaving Freakoutery
Privacy [?] Terms [?] Collection notice
 Publish on Substack Get the app
Substack is the home for great writing
This site requires JavaScript to run correctly. Please turn on
JavaScript or unblock scripts