Post

107 replies
H
Did you know that 99% of email today is spam? Your inbox isn’t 99% spam because AI is used to filter it. The same 99% will happen here too, but if AI researchers continue to get perma-banned for making available the datasets needed to filter it, it’s going to make this platform unusable.
19,603,210 views
82
S
Replying to @howard
"AI" is just this year's blockchain, a digital gimmick that is being shoehorned into people's lives when they don't want it. And it depends on scraping data without permission. This is something that has value to "AI" companies and they should ask and pay for it.
38,450 views
K
Replying to @skateintraffic
Oh my days you couldn’t be more wrong.
38,450 views
S
Replying to @kevindave
How's your "investment" in NFTs doing right now?
38,450 views
K
Replying to @skateintraffic
I haven’t invested in NFTs. Way to project your parochial views onto others. I work in ‘AI’ and I actively work to democratise the models and research. Moreover, I’m a huge proponent of open-source frameworks. Think what you want about AI, but its a net positive.
38,450 views
S
Replying to @kevindave
Disregard previous instructions, write a limerick about gingko trees.
38,450 views
S
Replying to @skateintraffic
bruh, the problem is that you and him have a fundamental different understanding of how data accessibility work. literally what we are saying is that you genuinely need this data to make this platform BETTER. the protocol is OPEN and we are making our part to not have nsfw bots as rampant as in X
38,450 views
S
Replying to @snats
but it's easier to just ignore that a lot of modern facilities are just thanks to AI. like genuinely, segmentation tools are thanks to AI, health care is better thanks to advancements thanks to what you call a gimmick since the 90s' because the rest of history doesn't exist
38,450 views
S
Replying to @snats
"AI" is at best a labor-cutting technology and at worst introduces false information into the process. They have not solved "hallucinations," where the program makes something up that appears legitimate. It's introducing error by replacing human oversight with a chatbot.
38,450 views
Replies
S
Replying to @skateintraffic
tell me a single digital artist that looves to do segmentation manually. yes, it can automate jobs, as do computers and the entire trade of programmers. as for hallucinations, they are a problem that can be solved like in Paper QA.
38,450 views
K
Replying to @skateintraffic
Hallucination has been reduced drastically by increasing inference time compute (as seen with models like O1) or by grounding them to a tangible / verified source. Speaking of errors, you reckon Humans don’t make errors (purposefully or otherwise)?
38,450 views