F
In the blue sky dataset drama, I see the pitchfork people using words such as "massive damage". I absolutely understand and support that someone has the right to control if their data is used or not to train AI systems. But what damage are they talking about?
22,390,900 views
Replies
P
O
Replying to
@francoisfleuret
Two guidelines I’ve adopted since moving here: • Don’t interact with anons • Don’t interact with righteous indignation This helps me avoid questions like these
11,500 views
C
Replying to
@francoisfleuret
There is no damage from curating a dataset. If(!) someone trains a model that causes damage on yhat dataset, then people should criticize the creator of that model. bsky.app/profile/chri...
11,500 views
A
Replying to
@francoisfleuret
People are currently reactionary towards AI. It's a consequential technology that extracts unpaid labor and is wielded by some powerful individuals in a way that is outside our control. Is there NEW "massive damage" resulting from this dataset? No. The damage of LLMs on society is already there.
11,500 views
J
Replying to
@francoisfleuret
it's a damage to the false conception that i think some artist might have had that this platform was like a safe-haven from that "evil corpo selling your data to train models and not giving you a cut" stuff. They dont know that with default the bluesky PDS ur data is 100% publicly viewable atm
11,500 views
M
Replying to
@francoisfleuret
The vitriol is disheartening. But the outrage says that the AI community should do better at addressing the growing distrust and aversion. Social media data has been used to cause "massive damage"(see Cambridge Analytica). The fact that some are failing to see the distinction btw that and beneficial
11,500 views
S
Replying to
@francoisfleuret
stop being disingenuous and don't vagueblog me like a highschooler. you're an adult, not a child. my post was clear about the damage, and you're behaving immaturely to do twitter-style dunking go back to twitter and keep my posts out of your mouth
11,500 views
A
M
D
Replying to
@francoisfleuret
It is especially awkward when the AT protocol docs highlight large scale search and algorithmic feeds as explicit goals. A million posts is barely even “large”. cf “the goal of having large scale search and algorithmic feeds.” (atproto.com/guides/faq)
11,500 views