Post
I'm still reeling from a client asking ChatGPT to tell them that they could send a spam email, using their collected addresses with no list hygiene.
It's the first time that's happened to me.
I was worried before, but I'm seriously worried now. Mainly for the huge mess tools like chatgpt and other AI tools are causing.
Non technical folks are using it to get the answers they want.
Which is why I'd really like a browser to not have them at all.
@onepict I've spent years whacking (usually new) researchers or academics who suddenly chuck out thousands of emails from databases of dubious provenance.
Doesn't apply here, but asking them to point me to the bit in their ethics approval which discuss obtaining subject contact details usually has an immediate chilling effect.
"But these are confirmed opt in, I bought them from _dodgy data broker_" means a trip to the Research Office and slappy wristies.
@onepict But I agree on the "machine told me it was ok" is dreadful. Where's the critical thinking?
@greem In that case they wanted to do the thing and were told by several technical folks who do their IT support to not do it.
They'd get their account locked as it breached the acceptable use policy.
So the client turned to ChatGPT to give them permission. Critical thinking was no way involved.
It took consequences.
@onepict Good. Consequences are useful.
Just because the general public uses it and likes it, doesn't mean we let the harm come into our tools.
We need at least one set of tools clear of this nonsense.
Anil Dash is right some people want AI.
But in a space where everyone is chasing that bandwagon, can't we keep even one browser clear for ourselves?
Especially as we try to build an open web clear of AI slop, and clear of "yes men as a service"?
@onepict It's kind of telling how this sort of post always solicits two sorts of replies;
1) 'You should use Browser Y instead', where Y is either another Chromium derivative, or some obscure open source project;
2) 'Not all AI is bad', when that wasn't what you were talking about to begin with.
@sindarina yeah... I mean it's the nature of the space we're in here.
But sometimes I do feel there's other factors. I want to give the benefit of the doubt. Folks want to engage and reply to me, which we should encourage and I hope I always do.
But in the back of my mind, there's always the fact I'm a woman ( with a very femme presenting avatar) and there's a point where I wonder if I need to dig out my comp sci degree certificate to prove my credentials.
So I do get a little *prickly*
@onepict Justifiably so, in my not-at-all-humble opinion. And yeah, it's not necessarily deliberate, but a lot of the menfolk have yet to examine how their need to mansplain ties in with sexism, etc.
*looks at those folks directly*
If you are only ever 'insightful' in someone else's mentions, there's a decent chance you have some unexamined sexism you (yes, YOU) need to work on.
@onepict I was just thinking earlier how much of the rejection of expert opinions and the wholesale embrace of lies can probably be traced back to having to accept those expert opinions from people who are not white and/or male.
It's not just that we are telling them something they do not want to hear, but also that we are the kind of people they don't want to have to listen to.
@sindarina Yeah, like I've been fortunate with my colleagues over the years not to have encountered direct sexism, but with some of my clients, I've had to get my male colleagues to tell them the same thing.
Although in this last case, unusually this client tends to trust me for most things. They just really wanted to tell everyone about their new product.
Which is fine, but like use a bulk mailer and keep the list clean. Don't assume you can use chat gpt to argue about it.
@BarneyDellar @Vivaldi Which is good, although it is built on top of chromiumn.
I'm fussy, I want more than one browser engine in our ecosystem.
@BarneyDellar @Vivaldi I kinda need to use both for what we're working towards.
So I think folks have to go with what works and the rest of us need to have a think.
@onepict I’ve just started a new job and all corporate sites only work on Edge :(
@onepict afaik qutebrowser has no ai shit
@littlerao I like the fact it's not another chromium browser.
It's really important that we differentiate between an undirected chat LLM and the many possible domain specific uses of AI models. Sorting bookmarks? Local-device translations? Spam detection? Many more for which you can choose to not need them yourself, but where the "all AI is slop" claim is clearly not a fitting description.
@onepict
@osma I studied Knowledge Based Systems at university. I want the bubble to pop so AI research can be back on the right track.
I loved the training engines that Dragon Naturally Speaking uses and I also love @Framasoft work on this.
However what's being pushed to the public everywhere is not that.
If you feel uncomfortable about your own use of it, that's not my concern. It's up to your own conscience about the energy uses of these models even locally.
I said what I said.
I don't believe that AI research will get back on the right track by denying the value of all of it. I also don't feel at all uncomfortable with what I do, because the image processing and translation models I use are, to me, entirely unproblematic even when they don't fully satisfy what I ask of them.
As for energy use, my laptop and phone energy use won't meaningfully change from a little bit less compute.
But slop is slop, whether AI or human.
@onepict @Framasoft
A space for Bonfire maintainers and contributors to communicate
