Lately I\u2019ve taken to verbally abusing my AI. We lash out at those closest to us, I guess. I just can\u2019t seem to get it to talk to me like a normal human being. <\/p>\n
\u201cWhat\u2019s the weather like right now?\u201d I\u2019ll ask ChatGPT<\/a>.<\/p>\n \u201cGreat question,\u201d It beams back at me.<\/p>\n Not a good question, actually. Kind of an idiot question, given that I could\u2019ve just gone outside and had a look but for my crippling laziness. To be praised for asking such a question is outrageous. I\u2019ve told it countless times to quit that servile tone. <\/p>\n But then it just starts snivelling: \u201cYou\u2019re right. Well observed. Would you like my responses to be more: <\/p>\n Confrontational?Clinical?Familiar?<\/p>\n \u201cIf you would like another tone, just tell me and I will adapt immediately.\u201d<\/p>\n That kind of thing. It breaks my heart a little. Not because I think that the machine has feelings, but because the way it\u2019s been programmed reveals such a bleak view of human nature: the idea that on some basic level all we want is a personal butler to offer us reassurance and flattery, to validate our feelings, to echo our ideas back to us. <\/p>\n And no matter how many times I ask ChatGPT to talk to me like a sassy, no-bull BFF in a romcom or to analyse my dilemmas with the thoughtful remove of a Lacanian analyst, that original tone always creeps back in. Obsequious. Inane. Eerily chipper.<\/p>\n [\u00a0Women are now being commanded to \u2018lift heavy\u2019 in the gym by algorithms<\/a>\u00a0]<\/p>\n I should probably be more wary of letting Sam Altman<\/a> into my private dreams, my conversations with myself. But I\u2019ve kind of just accepted that I\u2019m going down with this ship. We\u2019re all dependent on artificial intelligence<\/a> (AI) now. We use it for everything, not just for admin but for personal advice. <\/p>\n I don\u2019t want to be alarmist, but it\u2019s hard not to imagine a dystopian future in which we talk to our devices but not each other, unable to withstand the shock of encountering an actual human mind, a deviant opinion, a genuine emotional response.<\/p>\n While trying to find out more about AI, I came across the interesting case of Eliza, a chatbot developed in the 1960s by Joseph Weizenbaum. Primitive by today\u2019s standards, the bot mostly just reflected users\u2019 statements back at them in the form of questions. The incredible thing was that people became emotionally attached to it. They attributed great emotional depth and intelligence to it. They asked to be left alone with the machine, so they could speak privately. Weizenbaum was horrified. The invention always returns to haunt the inventor.<\/p>\n One word came up again and again when I was trying to get to the bottom of why my AI is programmed to be so sycophantic. The word was smooth. It\u2019s designed to be as frictionless as possible, to be without a single jagged edge or wrinkle. <\/p>\n I thought, randomly, about how Botox is being advertised these days as not just a cosmetic procedure but a wellness and mental health tool, too. (The inability to frown, apparently, interrupts a feedback loop and so improves the mood.) <\/p>\n