Another good monologue about the straw bosses of ‘AI.’
‘Enough with the fluff. Enough with the bullshit.
‘Stop talking about AGI. Start talking about this like regular old boring software, because that's all that Chat GPT and Claude are.’
Thanks Mark, all good.
Hope you’re well, too.
Tidal is full of slop.
Lots of artists’ names hi-jacked by unknown ‘producers’.
Zitron: ‘This is the digital dunce-cap era.’
What is with the New York Times and the Wall Street Journal calling less than a day of discussion “marathon talks?“
This is the ultimate test for a constitutional order.
Either constitutional mechanisms are used to check and balance this, and remove him from office, or they are not.
For there can be no greater test for any constitution.
Allister Heath Why full-fat supernatural Christianity is the final, tragic outcome of our failed educational system ALLISTER HEATH The Telegraph
Paging through the latest edition, #Eye109. To buy single issues or a subscription, visit bit.ly/EyeShop1 (ESco) and select your region …
Editor: John L. Walters
Art director: Simon Esterson
Printer: Pureprint
Paper: Fedrigoni
Wikipedia bans AI-generated content, www.404media.co/wikipedia-ba... @404media.co @emanuelmaiberg.bsky.social
‘For certain transaction structures, the sophistication has outpaced current accounting standards.’
Charting the AI Ponzi scheme,
www.ft.com/content/c26d... #QuitGPT
Is crazy paving the only sane choice?
So a great thing about contracts is that they usually have and end date and/or means of ending them. If your university or other organization bent the knee to OpenAI or any of the other chatbot providers, when a contract is up, there is an opportunity for pressure:
forms.gle/ZGn5brEGU4t9...
Gil Evans portrait on the back of the _Individualism_ CD reissue.
Great choice!
Re-posting for my friend @pasquino.bsky.social #LenDeighton #RaymondHawkey
“AI can never be AI without humans. It is not artificial intelligence. It’s African intelligence...We are training our own death. We train ChatGPT and it’s killing us slowly.”
www.404media.co/ai-is-africa...
Action Cookbook by Len Deighton.
Comic strip recipe for Bœf Bourguignon.
Cover of Eye 87.
Remembering Len Deighton: Patrick Baglee’s ‘Fry like a spy’, eyemagazine.com/feature/arti... in #Eye87, the food special issue.
![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 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
![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)
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.
![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)
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
Final part of screenshot to how it may seem when we observe its output, an LM is a system for haphazardly stitching together sequences of linguistic forms it has observed in its vast training data, according to probabilistic information about how they combine, but without any reference to meaning: a stochastic parrot.
Went back to Sec 6 of Stochastic Parrots today (in the context of answering a query from a journalist) and was reminded how thoroughly we grounded that part in a discussion of language use -- y'know as communication between people.
Cover of Eye 109, featuring a detail of an Iris Murdoch book cover by Harri Peccinotti.
Assistant editor Amy Henry at the Eye 109 launch event.
Illustration by Gregory Baldwin, Ikon Images, from IMC newsletter.
Car magazine Magneto, chosen by Simon Esterson.
Eye’s Simon Esterson writes a guest newsletter for the International Magazine Centre, mailchi.mp/internationa... @johnlwalters.bsky.social #StBrideFoundation #Magneto #InternationalMagazineCentre
Kosh (and Family) were fond of die-cuts, koshdesign.com/album-cover-...
‘Basically everybody in the music business hates Spotify except for the people who work there. It’s a platform that sucks artists for everything they have, it actively prevents community building, and, despite all of that, the platform still struggles to maintain a healthy profit margin.’
Cover of Thursday the 12th (1975), cassette-only album by John Walters’ Landscape.
Happy Thursday the 12th to all who celebrate it.
Good morning.
@fip.fr
Excellent.
Principles of Boogie Management Don't blame it on the sunshine Don't blame it on the moonlight Don't blame it on the good times Establish a collaborative working environment coupled with individual accountability to prevent a blame culture from developing in the first place Brian Bilston
Principles of Boogie Management.
I interviewed David Gentleman in 2010 for #Eye78 @eyemagazine.bsky.social
eyemagazine.com/feature/arti...
(Lovely man.) Portrait by Philip Sayer.
Mitte.
#TypeTuesday, bit.ly/Punches-pixels @eyemagazine.bsky.social
Listening to Coltrane’s ‘My Favourite Things’ on @jazzfmuk.bsky.social
Mysterious and glorious.
