The trouble is, unless an author rolls their own AI – perhaps a high bar for a free text-game competition – the player isn’t actually administering the Turing test, just trying to determine which bit of human-authored text is meant to denote personhood and which is meant to come from a machine intelligence. In a genre where text comes first, what better challenge than to closely read the responses of a mysterious interlocutor and separate out man from machine? And of course to have an AI sufficiently advanced for the test to be plausibly attempted almost requires a science-fictional setting of the type that tends to provide good fodder for a game, not to mention a likely-rogue robot or something to provide a readymade antagonist. It’s easy to see how the Turing test could be a good fit for IF. My son Henry was born right before the Comp, meaning I was fairly sleep-deprived and loopy while I played and reviewed many of the games, so in addition to a highlight and lowlight, the review includes an explanation of how new fatherhood has led me to betray the hard work the author put into their piece) All Rights Reserved.(This is a lightly-edited version of a review posted to the IntFict forums during the 2021 IFComp. Where are my clothes?” Its response: “I have a lot of clothes.”Ĭopyright ©2023 Dow Jones & Company, Inc. One example, published in August 2020 in MIT Technology Review: GPT-3 was asked, “Yesterday I dropped my clothes off at the dry cleaner’s and I have yet to pick them up. It doesn’t have common-sense reasoning or the ability to keep track of objects over time in a discussion. But at a global level-considering the meaning of multiple sentences, paragraphs or a back-and-forth dialogue-it becomes apparent that GPT-3 doesn’t understand what it’s talking about.
2).Ĭompared with earlier text-generation systems, the output generated by GPT-3 looks impressive at a local level individual phrases, sentences and paragraphs usually demonstrate good grammar and look like normal human-generated text. Eric Schmidt was executive chairman while I was in the trenches at Google in 2012, but I know better than to claim-as he does with Henry Kissinger and Daniel Huttenlocher-that GPT-3 is “producing original text that meets Alan Turing’s standard.” The GPT-3 program hasn’t passed the Turing test, and it seems nowhere near doing so (“ The Challenge of Being Human in the Age of AI,” op-ed, Nov.
0 kommentar(er)
