Neal Stephenson: evaluation

Transcript context

…In what year do you think we will first see a novel co-authored by a human and an AI, where the AI does more than just some basic research, but actually writes a fair amount of the novel, and that the novel is at least as good as what a talented human could accomplish? Something close to that has probably happened already. I’m sure it’s happened. So, your question boils down to a critical judgment as to quality. My position on this, which I put out in a Substack a couple of weeks ago, is that the real purpose of art and the reason we like art is because it exposes us to a very dense package of micro decisions that have been made by the artist. As such, we’re engaged in a communion with that artist. What makes it interesting is that connection. It may be to a living writer, or it may be to a sculptor who died 2,500 years ago, but in either case, we’re making a human-to-human connection. If we know that we’re reading something or experiencing a work of art that was just generated by an algorithm, then that element of human connection isn’t there anymore. Furthermore, we know that for every one such book or image or whatever, a million more could be generated just by launching a Python script, basically. Familiarity, the law of supply and demand takes over at that point. If there’s an infinite supply of this stuff, then its value drops to zero. Even if such a book reads pretty well as a novel, I don’t think people are going to be interested in it. But they may not know it was partly written by an AI or fully written by an AI, right?…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.