illustrata
RT @illustrata_ai: ✨I have two collections available rn.

On eth is The Flier - the 3rd edition piece in a series of 4.

On tez, Pastel Dreamscapes - a mix of small and medium editions & 1/1s🖤 https://t.co/04YTqLmNz5
tweet

GanWeaving 🐦
RT @arvalis: hahahah musk changed all the checkmark descriptions so people cant as easily tell if they are blue or legacy. https://t.co/UV0HKjf7ie
tweet

Yann LeCun
Many AI safety discussions today seem as speculative as discussions about airliner safety in 1890.

Before we have a basic design & basic demos of AI systems that could credibly reach human-level intelligence, arguments about their risks & safety mechanisms are premature. https://t.co/8ZHQbVOcEm
tweet

Rox
In conclusion I'll be implementing feature flags later this week lmao https://t.co/6lgcKh8iSd
accidentally just leaked a big new https://t.co/0KD2dhDOZs feature because I changed a hidden field to visible

then I tried to change it back and accidentally HID AN EXISTING FEATURE

so anyway if anyone tried to start a test in the past 20 minutes i am sorry - Rox
tweet

Rob Henderson
"people are often unaware of some of their own motivations....'If introspection were sufficient to reveal the true structure of human psychology, it’s not clear why we would even need to do science; we would just know.'" https://t.co/gP5DaQvzXK
tweet

Rox
accidentally just leaked a big new https://t.co/0KD2dhDOZs feature because I changed a hidden field to visible

then I tried to change it back and accidentally HID AN EXISTING FEATURE

so anyway if anyone tried to start a test in the past 20 minutes i am sorry
tweet

Firas Durri
Interesting to discover that the algorithm really pays attention to stuff like "did the poster engage with replies". So Twitter Inc was paying attention to this concept that some posters with 500k followers still read their replies & some just use Twitter as a syndication medium https://t.co/lOGvyFiZRw
Remarkable engagement rate differences in Twitter... CNN has 60 million followers, gets around 150k views, and a few dozen likes. My little personal profile 'converts' much better https://t.co/D1gpDHDYlD - Firas Durri
tweet

GanWeaving 🐦
RT @chetbff: I see you all out there. Hope all is going well.

Prompt:
Abstract Fractal circular mosaic city architecture made of millions of 3d odd shaped blocks, dot matrix background, as seen from the sky --q 2 --s 750 https://t.co/UdpP1ikq9B
tweet

GanWeaving 🐦
Do you want some whine with that, @elonmusk ? https://t.co/VtfOdSjX6s
@dogeofficialceo @nytimes NY Times is being incredible hypocritical here, as they are super aggressive about forcing everyone to pay *their* subscription - Elon Musk
tweet

Omar Khattab
RT @lateinteraction: ColBERT-PRF is one of coolest retrievers based on ColBERT out there.

It does Pseudo Relevance Feedback by expanding queries with *embeddings* obtained from the multi-vector representations of the retrieved documents, with strong results! https://t.co/V5l9mckpqI
@yoavgo In the era of neural IR, short user queries can still be hard to have a good representations in the dense retrieval paradigms, further enhancement brought by PRF evidence can improve this: https://t.co/IWGjYYDlCC, https://t.co/uDz22EEOnj - Xiao Wang
tweet

Yann LeCun
An oversimplification, but a rather accurate one.
Not clear to me that @geoffreyhinton is that far to the left of me. https://t.co/y6Cv0OZWq1
@ESYudkowsky @ylecun I've made a matrix so that people have a better understanding of "who stands where".

Probability of humans aligning AI vs probability of AI being hostile if not aligned by humans.

Please let me know what are your positions so it's not just my subjective perception.
Thanks https://t.co/D42KCRkgmp - Anton Gaia 🇺🇦
tweet