Show newer

Convo I had to have about using ChatGPT in our research 

Imagine that it does work exactly the way you want. Now you have invented a machine that converts clinical trial protocols into structured diagrams, but every time you use it, someone calls a Kenyan worker the n-word 100 times. How do you feel about your invention?

TIL: there are languages that have inclusive and exclusive forms of first person plural.

Like, you can say "we" in a way that means "we with you" or "we without you", which is a semi-common thing to happen in e.g. English and German.

"we're going to have to fix this" – "wait, we as in we or we as in you?"

Mandarin for instance:

wǒmen (us, but not you)
zánmen (us, including you)

PSA: Enable browser.urlbar.suggest.calculator in your firefox about:config so typing math expressions in the address bar shows the result with an option to copy

meta, blocklists 

A way in which mechanically shared blocklists could *reduce* incorrect listings (if implemented correctly): retractions can propagate or at least notify automatically, rather than having to manually distribute updated information and hope that it reaches everyone who originally blocked.

(Not just for cases where the block recommendation turned out to be a mistake, but more importantly for instances that were correctly blocked but have cleaned up their act over time)

genitalia mentioned 

so I made a test block called testicle_block, just to see how different it was registering blocks between 1.20 and 1.7.10

If I renyame it to anything else it breaks.

I am stuck with a testicle_block

#OpenAI is so dedicated to AI safety that as far as I can see, they don't even have a way to report misuse which contravenes their Usage Policies openai.com/policies/usage-poli

You may not realise because you may not be looking at the bigger picture, but there are people from what we call 'alt fedi', which is where the racist channer shitheads hang out, who love any excuse to bash on Black folk, particularly Ro because he outs them constantly, and they see you all shit-talking The Bad Space and Ro and they love it. They LOVE that shit. So they push it, and you buy into it because right now it suits your narrative.

It is toxic and you are feeding the racist trolls.

Show thread
@ipg tell you one better: windows programs running in wine can execute linux syscalls with no restrictions

Even though I rationally understand how it works, it still feels weird to see "aged 12+ months, consume within 5 days" on a block of cheese

I'm testing my reach here. Probably won't work, but do these messages reach anybody currently living in Belgium? I have an odd project that requires "boots on the ground" near Brussels... #belgium #belgique #brussels

re: meta meta, blocklists 

(I'd invoke the concept of lateral violence here, but in this case it's a bit more complicated because getting the approach to blocklists wrong also risks causing *actual* additional harm)

Show thread

@welshpixie Running a conspiracy is for neurotypicals whose lives have gone wrong, everyone else is busy or tired.

meta meta, blocklists 

It occurs to me that a lot of the discourse around blocklists can get extremely heated without ever talking about the role of the malicious instances *getting* blocked.

That doesn't mean that there aren't legitimate concerns about the usage of shared blocklists or lack thereof, but when folks become angrier at folks discussing the merits of blocklists than at the shitty people who necessitate all the blocking in the first place... that doesn't seem like a desirable outcome.

the arguments there must be really funny

­>"no this fighter jet has a third 36-J engine"
­>"nuh-uh"
­>
102856428-MITSUBA6-CL.pdf
­>"yuh-uh"

Show thread

Well, today sure has been A Day.

(Most of it wasn't on fedi)

Mac OS file manager: What, you didn't want me to vomit .DS_Store files all over everything?

Linux file manager: What, you wanted to choose your own bookmarks?

Windows file manager: What, you didn't want advertisements in your file manager?

Periodic reminder for y'all kids, teens & 20-somethings 'growing up gifted’: find and do something you have a legit chance of failing at. Ideally, a few times. Practice failing.

If you coast through life because it's comparatively easy (you're gifted!! yay!) you will not have a useful mental context when failure inevitably happens, and that's Bad!

(There's a corollary in here about self-worth/esteem vs effort expended, but maybe I'll save that for a blog post.)

Show thread
A common argument I come across when talking about ethics in AI is that it's just a tool, and like any tool it can be used for good or for evil. One familiar declaration is this one: "It's really no different from a hammer". I was compelled to make a poster to address these claims. Steal it, share it, print it and use it where you see fit.

https://axbom.com/hammer-ai/

#AiEthics #DigitalEthics
Show older
Pixietown

Small server part of the pixie.town infrastructure. Registration is closed.