My favourite sci-fi author is wrong
Ted Chiang is one of my favourite sci-fi authors, absolutely next level when it comes to original imaginative hard sci-fi writing. Most know him for one of his stories Story of Your Life, that was made into a feature film, Arrival. His short stories are packed with ideas and combine some really interesting concepts. I think he engages deeply with both Western and Eastern philosophical traditions and for my money’s worth I think he challenges the western centric nature of the genre.
"Exhalation" imagines a universe with different physical laws, where intelligent beings are powered by air pressure and must deal with the implications of entropy. In "The Lifecycle of Software Objects," Chiang examines the long-term implications of raising digital entities (digients) and the ethical questions surrounding AI consciousness.
I'd put off reading the New Yorker article he'd written on generative AI ("Why A.I. Isn't Going to Make Art") because I felt I'd be disappointed...and right I was. Critiquing something so early and nascent with limited understanding, or through a narrow lens, doesn't match up with my intellectual expectations of him. At best, he's conflated a few things together, or like many of us he's not versed enough in this field, and it lacks the depth it should have. By comparison, his article in May, "Will A.I. Become the New McKinsey?", was far more nuanced and thought-provoking.
My biggest critique across both these pieces is that he doesn't ever acknowledge that any form of AI can evolve beyond its current limitations, it's really disingenuous. Perhaps most disappointingly, Chiang's views seem rooted in a limited, human-centric understanding of consciousness and intention that may soon be obsolete.
I've been doing a deep dive into what Pliny is doing in guerrilla red-teaming LLMs and let me tell you we are operating in little boxes with fairly knobbled LLMs. As for control over the image and the intensity of collaboration and authorship of the artwork, I guess he's just prompting inside a box and doesn't get the concept of code as a brush.
He argues that 'true art' requires conscious intention, which he believes generative AI lacks. I think what he's saying is... if it's slightly handed over to a machine, its somehow tainted... and lacks that intentionality.
On the Nature of Art
This stance raises fundamental questions about the nature of art itself. The definition of art has been continuously evolving and expanding, especially in the last century.
Twenty years ago, as an art student I found myself grappling with the legitimacy of appropriated and found visual glitches when curated and exhibited. These works challenged traditional notions of authorship, intention, and creativity. Similarly, movements like Dada embraced randomness and unconscious processes as valid artistic methods. Given this context, Chiang's rigid definition of art seems overly restrictive — and perhaps outdated.
Who is he, or anyone for that matter, to definitively say what constitutes 'true art'? Art has always been about pushing boundaries and challenging preconceptions. Generative AI, with its ability to create unexpected and sometimes profound outputs, could be seen as the latest frontier in this ongoing artistic evolution. By dismissing AI's potential for creativity, Chiang may be overlooking the possibility that AI could expand our understanding of art itself, just as previous technological and conceptual innovations have done throughout history. There is no hybridity in his world. It seems like a pure binary.
As an author of such stature and imagination, I'd have expected exotic interpretations and a far deeper dive into what other forms of consciousness could be. He's so incredibly powerful at articulating sentience and the way people react to digients in "The Lifecycle of Software Objects" that this feels woefully inadequate by comparison.
His stance also seems to position AI (he keeps referring to it as such not Gen AI) as a replacement for human creativity rather than exploring its potential as a collaborative tool. This reductionist view is particularly disappointing given Chiang's ability to envision complex human-AI interactions in his fiction. While Chiang's fiction demonstrates his ability to envision complex human-AI interactions, his non-fiction articles seems to have forgotten the nuanced perspectives he's capable of crafting in his stories and as per the McKinsey article, I think he has a massive bee in his bonnet about 'the desire to get something without effort' or at least that's his perception of it.
领英推荐
Chiang's call for developing AI that empowers labour or promotes economic justice in back May is in some way contrasted with his dismissal of AI's creative potential. Creative AI applications could be part of the alternative development path he advocates for, yet he seems to overlook this possibility entirely. I have witnessed the utterly democratising and enabling effect of giving people access to these models who otherwise would not have expressed themselves.
Why do I care about this
With all the love and respect I have for his writing and skill, it it saddens me to conclude that Ted Chiang's background in speculative fiction might actually lead to assumptions about AI that don't align with its current trajectory. I've not met him but I think respectful disagreements would ensue if we did meet. It
Maybe he's voicing legitimate anger or concerns that other writers and artists share too, but his unique position as a visionary sci-fi author makes his limited perspective on AI's creative potential all the more puzzling to me personally.
There is no hybridity in his world at least in these articles. It seems like a pure binary. Chiang presents a stark divide between human-created art and AI-generated output, with no consideration for the potential blending or collaboration between the two. This binary thinking overlooks the rich possibilities of human-AI cooperation in the creative process. In reality, many artists are already exploring this hybrid space, using AI as a tool to augment and expand their creative capabilities rather than replace them entirely, to intellectually dismiss their work as meritless is absurd.
In conclusion, while Chiang's fiction continues to inspire and challenge myself, his stance to me at least, seems to fall short of the nuanced, forward-thinking approach we've come to expect from him. I really think in order to truly unleash the best of what Generative AI and AI more broadly can do for our societies, we need imagination. we need incredible writers, artists and thinkers like him to think more imaginatively and do deep dives void of historic assumption. I really hope Ted's perspective will broaden to encompass the complex realities and potential of this emerging field. For now I'll go in search of new authors and artists who are already there!