Recent headlines have made clear: If AI is doing an impressively good job at a human task, there\u2019s a good chance that the task is actually being done by a human.<\/p>\n\n\n\n
When George Carlin\u2019s estate sued the creators of a podcast who said they used AI to create a standup routine in the late comedian\u2019s style, the podcasters claimed that the script had actually been generated by a human<\/a> named Chad. (The two sides recently settled the suit.) <\/p>\n\n\n\n A company making AI-powered voice interfaces for fast-food drive-thrus can only complete 30% of jobs<\/a> without the help of a human reviewing its work.<\/p>\n\n\n\n Amazon is dropping<\/a> its automated \u201cJust Walk Out\u201d checkout systems from new stores \u2013 a system that relied on far more human verification<\/a> than it was hoping for.<\/p>\n\n\n\n We\u2019ve seen this before \u2013 though it may already be lost to Silicon Valley\u2019s pathologically short memory. <\/p>\n\n\n\n Back in 2015, AI chatbots were the<\/em> hot thing. Tech giants and startups alike pitched them as always-available, always-chipper, always-reliable assistants. One startup, x.ai, advertised an AI assistant who could read your emails and schedule your meetings. Another, GoButler, offered to book your flights or order your fries through a delivery app. Facebook also tested a do-anything concierge service called M, which could answer seemingly any question, do almost any task, and draw you pictures<\/a> on demand. <\/p>\n\n\n\n But for all of those services, the \u201cAI assistant\u201d was often just a person. Back in 2016, I wrote<\/a> a story about this and interviewed workers whose job it was to be the human hiding behind the bot, making sure the bot never made a mistake or spoke nonsense.<\/p>\n\n\n\n To power Facebook\u2019s M, every single message was reviewed by contractors who worked out of the company\u2019s Menlo Park headquarters. The workers at GoButler and x.ai<\/a> told me they worked long hours or overnight shifts to mimic the bots’ perma-online presence. Customers constantly asked them if they were bot or human. Some customers sent them crude sexual messages, thinking no human would read it.<\/p>\n\n\n\n (In another indicator of Silicon Valley\u2019s rapid evolutionary cycle, x.ai closed in 2021, and the name is now attached to xAI, Elon Musk\u2019s new AI venture.)<\/p>\n\n\n\n People build AI to mimic human intelligence and capabilities. But when the AI can\u2019t quite deliver on the promise, we end up with humans pretending to be chatbots pretending to be humans.<\/p>\n\n\n\n It\u2019s the latest iteration of a trick that stretches back at least as far as 1770, when the original Mechanical Turk machine appeared to play chess automatically \u2014 but actually concealed a human chessmaster inside its apparatus.<\/p>\n\n\n\n Of course, some things are different this time around. Chatbots are more fluent in their conversation and mimicry. But today\u2019s bots, just like 2016\u2019s, struggle with reliability. We still trust a human more when something has to be done right.<\/p>\n\n\n\n Investor frenzy is often driving this hype. In 2016, \u201cHuman-assisted AI is \u2018the hottest space to be in right now,\u2019\u201d one founder said. Another founder said that she was frustrated with the way a lot of investors would \u201ccongratulate us on having a fully automated bot\u201d when the technology was quite reliant on humans. <\/p>\n\n\n\n As long as there\u2019s the incentive to overhype AI\u2019s abilities, there will be gaps between what AI promises and what it can reliably do. To fill that gap, you can always hire a person. \u2014Ellen Huet<\/a><\/p>\n","protected":false},"excerpt":{"rendered":" Recent headlines have made clear: If AI is doing an impressively good job at a human task, there\u2019s a good chance that the task is actually being done by a human. When George Carlin\u2019s estate sued the creators of a podcast who said they used AI to create a standup routine in the late comedian\u2019s style, the podcasters claimed … Continue reading “Bloomberg TechDaily: Humans imitating computers imitating humans”<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4],"tags":[],"class_list":["post-4096","post","type-post","status-publish","format-standard","hentry","category-social"],"_links":{"self":[{"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/posts\/4096","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4096"}],"version-history":[{"count":1,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/posts\/4096\/revisions"}],"predecessor-version":[{"id":4097,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=\/wp\/v2\/posts\/4096\/revisions\/4097"}],"wp:attachment":[{"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4096"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4096"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/softbeam.net\/hobby\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4096"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}