Reposted by Srinivas Inguva
And now, the video of “Darth Vader”on why he wants Flock in San Diego.
Read the full story here: www.404media.co/darth-vader-...
Srinivas Inguva
@paranoidbaker.bsky.social
217 followers 253 following 57 posts
paranoidbaker @ {twitter, macaw.social, bsky, threads, t2.social, spill}. Security, books, and snark.. used to work for the bird.
![[alt text by NASA] The Moon, seen here backlit by the Sun during a solar eclipse on April 6, 2026, is photographed by one of the cameras on the Orion spacecraft’s solar array wings. Orion is visible in the foreground on the left. Earth is reflecting sunlight at the left edge of the Moon, which is slightly brighter than the rest of the disk. The bright spot visible just below the Moon’s bottom right edge is Saturn. Beyond that, the bright spot at the right edge of the image is Mars. Credit: NASA](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:62gdkmpwa7xme36getyqnem3/bafkreiakufyngycqogzws43gzq5ei2zayvoozx42wmbmyawywxyjqhw2ua)
![Earthset captured through the Orion spacecraft window at 6:41 p.m. EDT, April 6, 2026, during the Artemis II crew’s flyby of the Moon. A muted blue Earth with bright white clouds sets behind the cratered lunar surface. The dark portion of Earth is experiencing nighttime. On Earth’s day side, swirling clouds are visible over the Australia and Oceania region. In the foreground, Ohm crater has terraced edges and a flat floor interrupted by central peaks. Central peaks form in complex craters when the lunar surface, liquefied on impact, splashes upwards during the crater’s formation.
[alt text from NASA]](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:62gdkmpwa7xme36getyqnem3/bafkreid6rp25tpuxaceaziwiduvtsnbystj5s542tlmuhcstyuxyfaj7ve)
![Screenshot from paper:
6.1 Coherence in the Eye of the Beholder
Where traditional n-gram LMs [117] can only model relatively local dependencies, predicting each word given the preceding sequence of N words (usually 5 or fewer), the Transformer LMs capture much larger windows and can produce text that is seemingly not only fluent but also coherent even over paragraphs. For example, McGuffie and Newhouse [80] prompted GPT-3 with the text in bold in Figure 1, and it produced the rest of the text, including the Q&A format.21 This example illustrates GPT-3’s ability to produce coherent and on-topic text; the topic is connected to McGuffie and Newhouse’s study of GPT-3 in the context of extremism, discussed below.
We say seemingly coherent because coherence is in fact in the eye of the beholder. Our human understanding of coherence de- rives from our ability to recognize interlocutors’ beliefs [30, 31] and intentions [23, 33] within context [32]. That is, human language use](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:aeug3w4mdhnc2wv73hmrh36v/bafkreidfgm4ycx3q6rcfqinabje5bstgky6va5q6kto2isges7ry2cn54u)
![Screenshot continued:
takes place between individuals who share common ground and are mutually aware of that sharing (and its extent), who have commu- nicative intents which they use language to convey, and who model each others’ mental states as they communicate. As such, human communication relies on the interpretation of implicit meaning conveyed between individuals. The fact that human-human com- munication is a jointly constructed activity [29, 128] is most clearly true in co-situated spoken or signed communication, but we use the same facilities for producing language that is intended for au- diences not co-present with us (readers, listeners, watchers at a distance in time or space) and in interpreting such language when we encounter it. It must follow that even when we don’t know the person who generated the language we are interpreting, we build a partial model of who they are and what common ground we think they share with us, and use this in interpreting their words.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:aeug3w4mdhnc2wv73hmrh36v/bafkreiciu4ell2f66q3dxjuk65gk2leau3rfxlaylbk3mklnqrf6gxo3sm)
![Screen shot continued
Text generated by an LM is not grounded in communicative intent, any model of the world, or any model of the reader’s state of mind. It can’t have been, because the training data never in- cluded sharing thoughts with a listener, nor does the machine have the ability to do that. This can seem counter-intuitive given the increasingly fluent qualities of automatically generated text, but we have to account for the fact that our perception of natural language text, regardless of how it was generated, is mediated by our own linguistic competence and our predisposition to interpret commu- nicative acts as conveying coherent meaning and intent, whether or not they do [89, 140]. The problem is, if one side of the commu- nication does not have meaning, then the comprehension of the implicit meaning is an illusion arising from our singular human understanding of language (independent of the model).22 Contrary
fn22: Controlled generation, where an LM is deployed within a larger system that guides its generation of output to certain styles or topics [e.g. 147, 151, 158], is not the same thing as communicative intent. One clear way to distinguish the two is to ask whether](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:aeug3w4mdhnc2wv73hmrh36v/bafkreid2ta4cavxeinzouycdyc57jvu4xcvlew7gfbn4s554ip3totwpaa)
