Discussion

• @jathansadowski Jathan Sadowski on x

  As much as Ted Chiang's fiction is fantastic, it's really his non-fiction that I could read forever. His two paragraphs here about using ChatGPT as a writing tool are the most thoughtful comments I've seen about its utility and the practice of writing. https://www.newyorker.com/.…

• @ryanqnorth Ryan North on x

  I've been waiting for someone to write an article like this for a while! It gracefully and accessibly explains, from a computer science perspective, what “AI” like Chat-GPT is doing, what it's NOT doing, and where the value is. Worth a read! https://www.newyorker.com/...

• @adrianhon Adrian Hon on x

  Goddammit Ted Chiang, is it not enough for you to be an amazing fiction writer that you also have to defeat us all at non-fiction? https://www.newyorker.com/...

• @papapishu @papapishu on x

  I think there is a lot to this analogy. AI as a series of lossy copies that slowly degrade over time. https://www.newyorker.com/...

• @jathansadowski Jathan Sadowski on x

  Of course Ted Chiang would provide one of the best, clearest descriptions of how exactly ChatGPT works. Explaining that large language models are essentially a hyper compressed image of the entire web is a clever analogy that makes perfect sense. https://www.newyorker.com/... htt…

• @neilturkewitz Neil Turkewitz on x

  A brilliant piece from Ted Chiang in ⁦@NewYorker⁩: “Obviously, no one can speak for all writers, but let me make the argument that starting with a blurry copy of unoriginal work isn't a good way to create original work.” https://www.newyorker.com/...

• @odannyboy Dan Saffer on x

  “Most of our fears or anxieties about technology are best understood as fears or anxiety about how capitalism will use technology against us.” Ted Chiang on ChatGPT https://www.newyorker.com/...

• @mdbell79 Matt Bell on x

  Ted Chiang on the “hallucinations” of ChatGPT: “if a compression algorithm is designed to reconstruct text after 99% of the original has been discarded, we should expect that significant portions of what it generates will be entirely fabricated...” https://www.newyorker.com/...

• @henrycooke Henry Cooke on x

  This Ted Chiang piece on ChatGPT as a lossy compression algorithm for the web is the best thing I've read on it as a tool for writing. https://www.newyorker.com/... https://t.co/...

• @gopalsathe Gopal Sathe on x

  “It retains much of the information on the Web, in the same way that a jpeg retains much of the information of a higher-resolution image, but, if you're looking for an exact sequence of bits, you won't find it; all you will ever get is an approximation.” https://www.newyorker.com…

• @simonkirby Simon Kirby on x

  Chiang on resisting GPT as writing tool:"Your first draft isn't an unoriginal idea expressed clearly; it's an original idea expressed poorly, and it is accompanied by your amorphous dissatisfaction, your awareness of the distance between what it says and what you want it to say."…

• @timfinin Tim Finin on x

  Award-winning science fiction author Ted Chiang (who has a CS degree, btw) has a new article in the New Yorker on ChatGPT. He makes an interesting analogy for what it does to lossy compression of the text from which its LLM was created. #AI #LLM #ChatGPT https://www.newyorker.com…

• @mraginsky Maxim Raginsky on x

  A thread on Ted Chiang's New Yorker piece about ChatGPT. WARNING: this will get into technical detail about lossy compression, generative modeling, and the meaning of the word “blurry.” 1/16 https://www.newyorker.com/...

• @tonyzador Tony Zador on x

  Brilliant discussion of ChatGPT as lossy compression ("blurry jpeg") of the internet, and why its lossiness contributes to it seeming so clever. https://www.newyorker.com/... https://twitter.com/...

• @jjvincent James Vincent on x

  ted chiang explaining the mechanisms that animate AI language models as a form of lossy compression is, of course, a perfect analogy — intuitive, simple, and illuminating https://www.newyorker.com/... https://twitter.com/...

• @gralefrit Joel Morris on x

  This is a brilliant point. AI isn't simulating human creativity, because it's not shit enough. Being creative isn't about doing stuff well, it's about doing stuff really badly, and fixing it. The only authentic simulation of human creativity in AI art is the fingers. https://twit…

• @pavelasamsonov @pavelasamsonov on x

  GPT is a blurry jpeg of the web, but in text compression artifacts seem like understanding. “When we're dealing with sequences of words, lossy compression looks smarter than lossless compression.” https://www.newyorker.com/...

• @katyilonka @katyilonka on x

  among other things, ted chiang's essay about chatGPT as a blurry jpeg of the web is a great example of the power of metaphor as a thinking tool: by thinking through the ways in which this metaphor is and isn't accurate, https://www.newyorker.com/...

• @brianroemmele Brian Roemmele on x

  In 100 years history will be generated in demand through a frosted window of reiteration of reiteration of reiteration of a copy from snippets. https://twitter.com/...

• @jjvincent James Vincent on x

  also this point is very telling: if OpenAI and others exclude AI-generated text from the training data of future LLMs, what does that tell you about the quality of that AI-generated text? https://twitter.com/...

• @michaelluo Michael Luo on x

  Really clarifying assessment by the great science fiction writer Ted Chiang. “There's nothing magical or mystical about writing, but it involves more than placing an existing document on an unreliable photocopier and pressing the Print button.” https://www.newyorker.com/...

• @thisishannahkim Hannah Kim on x

  “Starting with a blurry copy of unoriginal work isn't a good way to create original work... Sometimes it's only in the process of writing that you discover your original idea... Your first draft isn't an unoriginal idea expressed clearly; it's an original idea expressed poorly.” …

• @craigmod Craig Mod on x

  “ChatGPT is so good at this form of interpolation that people find it entertaining: they've discovered a “blur” tool for paragraphs instead of photos, and are having a blast playing with it.” https://www.newyorker.com/...

• @meadwriter K.C. Mead-Brewer on x

  “the more that text generated by large-language models gets published on the Web, the more the Web becomes a blurrier version of itself” god, I love Ted Chiang; fantastic essay https://www.newyorker.com/...

• @gwillowwilson G. Willow Wilson on x

  Ted Chiang is exactly the person you want to hear explaining AI. https://www.newyorker.com/...