최근 Replit과 같은 '바이브 코딩(vibe coding)' 스타트업들이 틱톡(TikTok) 인플루언서를 통해 소프트웨어 개발로 쉽게 부를 이룰 수 있다는 거짓된 희망을 팔고 있습니다. 이는 1920년대 다단계 마케팅(Multi-Level Marketing)이나 암호화폐 붐을 연상케 하는 구조로, 경제적 불안감을 느끼는 젊은 층의 절망을 착취하는 일환이라는 비판을 받습니다.
번역된 본문
생성형 AI가 허벌라이프를 닮아가는 이유: 기만적인 바이브 코딩 스타트업들이 청년층에게 거짓된 희망을 팔고 있는 방식
매튜 휴즈 (Matthew Hughes), 2026년 6월 17일
지난 몇 주 동안, 저는 틱톡(TikTok)에서 Replit의 광고를 자주 보았습니다. Replit은 지난 몇 년간 등장한 수많은 '바이브 코딩(vibe coding)' 스타트업 중 하나로, Anthropic, OpenAI, 구글(Google)의 모델을 그럴싸하게 포장해 놓은(glorified wrapper) 프론트엔드에 불과합니다.
때로 틱톡은 꽤 기이한 곳이기도 합니다. 절대 구매할 의사도, 능력도 없는 사람들에게 물건을 팔려고 시도하는 기업들로 가득 차 있으니까요. 산업용 글리신, 흑연 큐브, 심지어 중고 유조선 같은 것들이죠. 저는 보통 그런 광고들을 보며 웃어넘기고 일상을 이어갑니다. 비교해 보면, 소련 RBMK 원자로에나 어울릴 법한 흑연 큐브를 초연결 세대(Z세대)에게 팔려는 생각만큼 Replit의 광고가 미친 듯이 말도 안 되는 것은 아닙니다.
(What We Lost는 독자들의 지원으로 운영됩니다. 새로운 글을 받아보고 제 작업을 응원하고 싶으시다면 무료 또는 유료 구독자가 되어주세요.)
인플루언서들의 협찬 광고로 만들어진 이 영상들은(“replit”과 “#ad”, 또는 “#replitpartner”를 검색하면 볼 수 있습니다) 대체로 비슷한 패턴을 따릅니다. 매력적인 외모의 누군가가 일상을 보내다가 아이디어를 떠올리고, Replit을 사용해 프롬프트를 입력해 그것을 실제 앱으로 만들어내는 식입니다. 전달되는 메시지는 간단합니다. 바로 당신도 소프트웨어를 만들 수 있으며, 그 소프트웨어가 당신의 삶을 더 낫고 편리하게 만들어 줄 수 있다는 것입니다. 혹은 그 소프트웨어가 당신의 월세를 보태주는 훌륭한 부업(N잡)이 되어줄 수도 있고, 운이 좋다면 부자로 만들어 줄 수도 있다는 것이죠. 우리는 이 어디서 본 적이 있던 말들을 들어본 적이 있을까요?
역사는 되풀이된다
1920년대 무렵, 세계는 다단계 마케팅(MLM)이라는 유해하고 혐오스러운 개념을 접하게 되었습니다. 작동 방식은 간단합니다. 건강 보조제나 플라스틱 식품 용기 같은 제품을 가진 회사가 일반인들을 마케터이자 영업사원으로 끌어들이는 것입니다. 실제 제품을 팔아서 남는 돈은 거의 없습니다. 진짜 큰돈을 벌려면 다른 영업사원을 모집해야만 합니다. '다운라인(Downline)'이라 불리는 각 신규 모집자는 보통 일회성 보너스를 가져다주며, 그들이 판매하는 제품의 수익 일부를 떼어줍니다. 당연히 이제 그 다운라인들도 자신들만의 다운라인을 모집해야 합니다. 그리고 3세대 다운라인 역시 또 다른 다운라인을 모집해야 하죠. 이것이 지속 가능할 리 없습니다. 결국 사람이 바닥나고 맙니다. 일종의 피라미드 사기와도 같습니다.
이후 한 세기 동안 다단계 마케팅 산업은 변화했습니다. 글로벌화되었죠. 우리는 건강에 도움이 되는지 의심스러운 웰니스 음료나, 머리를 빠지게 만드는 샴풸 등을 파는 새로운 MLM 기업들이 우후죽순 겉불어나는 것을 보았습니다. 하지만 (애초에 중요한 것이 아니었던) 제품이 바뀌었을 뿐, 근본적인 작동 모델은 변하지 않았습니다. 사람들이 이 곳에 가입하는 근본적인 동기도 마찬가지입니다.
MLM 사기에 빠지는 사람들이 결코 부자가 아니라는 사실을 눈치채셨나요? 항상 재기가 불가능할 정도로 경제적으로 궁핍한 사람들입니다. 그래서 (적어도 미국에서는) 공식 경제 시스템에서 배제된 불법 체류 이민자들이 이 사기에 많이 가입합니다. 터퍼웨어나 단백질 쉐이크, 혹은 그 회사의 제품에 열정이 있어서 MLM에 가입하는 사람은 아무도 없습니다. 그들은 충분히 노력하면 자신들을 외면했던 경제적 안정감을 찾을 수 있고, 어쩌면 그 이상을 얻을 수 있다고 믿기 때문에 가입하는 것입니다.
물론, 실제로 가입한 사람 중 극소수만이 경제적 안정을 이루고, '그 이상'을 얻는 사람은 그보다도 더 적습니다. 시대가 어렵고 사람들이 경제 체제로부터 소외감을 느끼는 한, 항상 그 소외감을 악용하려는 무리가 나타나기 마련입니다. 세상에, 그저 2010년대 암호화폐 붐을 보기나 하십시오. 제가 The Next Web에서 일하던 시기는 그 시대의 거품이 가장 심하고 가장 멍청했던 시기와 정확히 겹쳤습니다. 당시 저는 매일 약 250개의 이메일을 받았는데, 창업자와 홍보 담당자들이 클라이언트나 회사에 대한 기사를 써달라고 졸라대는 내용이었습니다.
Generative AI Is Having Its Herbalife Moment How predatory vibe coding startups are selling false hope to young people Matthew Hughes Jun 17, 2026 50 19 12 Share Over the past few weeks, I’ve noticed that TikTok has been serving me ads for Replit — one of the many vibe-coding startups that have emerged in the past couple of years, and that serve as a glorified wrapper for models from Anthropic, OpenAI, and Google. Now, TikTok can be a weird place at times. It’s full of companies trying to sell things to an audience that can not — and will not — ever buy them. Things like industrial-grade glycine , graphite cubes, and lightly-used oil tankers. I usually just laugh at those ads and move on with my day. And, in comparison, Replit isn’t as outlandishly batshit as the idea of someone trying to sell a graphite cube — the kind that looks like it might feel at home in a Soviet RBMK nuclear reactor — to a bunch of hyper-online gen-Z. What We Lost is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Subscribe The ads — many of which were produced by influencers with brand deals, and can be found by searching “replit” and “#ad,” or #replitpartner — typically follow the same format. You have a beautiful person going about their day, having an idea, and then prompting it into existence with Replit. The message is simple — you, yes you, can make software, and that software might make your life some level of better, or more convenient. Or, that software might become the side-hustle that helps pay your rent, or even, if you’re lucky enough, makes you rich. Where have we heard this before? History Rhymes Around the 1920s, the world was introduced to the noxious and hateful idea of multi-level marketing. The way it works is simple — a company will have a product, like health supplements or plastic food containers, and they’ll get ordinary people to act as their marketers and salespeople. There’s very little money in actually selling product. If you want to make real cash, you need to recruit other salespeople. Every new recruit — often referred to as the “downline” typically brings in a one-time bonus, as well as a cut of every bit of product they sell. Obviously, those downlines now need their own downlines. And those third-generation downlines need their own downlines. Obviously, this isn’t sustainable. Eventually, you run out of people. Kind of like a pyramid scheme. Over the following century, the multi-level marketing industry changed. It globalized. We saw the emergence of new MLM companies spruiking wellness drinks with dubious health benefits, shampoos that make you go bald, and more. But while the product (which was never the point) changed, the basic model didn’t. Neither did the incentives for actually joining them. Have you noticed that it’s never rich people that join MLM schemes? It’s always those on the bones of their arse. It’s why a lot of the people joining them are (at least, in the US) undocumented migrants who are otherwise locked out of the formal economy. Nobody joins an MLM because they’re passionate about fucking tupperware or protein shakes, or the company manufacturing them. They join them because they believe that if they work hard enough, they can find the economic stability that otherwise eluded them, and perhaps something more. Of course, only a small percentage of people who actually join them actually achieve that economic stability, and a smaller number still that “something more.” So long as times are tough — that there’s people who feel alienated from the economic system — there will be people looking to take advantage of that alienation. Jesus, just look at the crypto boom of the 2010s. I had a front-row seat for much of it, with my time at The Next Web coinciding with the frothiest, dumbest part of the era. Each day, I’d get around 250 pitches from founders and PR people in my inbox, all asking me to write about their clients or companies, and I’d wager that around 200 of them (or four-fifths) had something to do with crypto. Tell me, do you think that the crypto boom would have happened if not for the fact that people saw a material decline in their economic fortunes and living standards after the global financial crisis, and that they didn’t see any hope of that changing? FTX pitched itself to a mainstream audience, buying the naming rights to sports stadiums and spending big on a marketing campaign that targeted normal people, with Superbowl ads fronted with recognizable celebrities like Larry David . Do you think that mainstream appeal would have been there — or would have been quite as strong — if not for the fact that people were broke, with their spending power eroded year-after-year as inflation chipped away at their stagnant wages? Would Paris Hilton have been invited to talk about NFTs on the Jimmy Fallon show , if not for the fact that the dire economic situation had created an opening for NFTs in the first place? The Vibe Coding Scam Things are fucked. More fucked than they’ve been in a long, long time. What makes this moment even more dangerous than, say, the pandemic era, or the period after the global financial crisis, is that at least people believed that this too shall pass. That with vaccines and natural immunity, and sensible policy decisions, life will return to normal. That eventually, the banks will right themselves. Right now, people — especially young people — are looking down the barrel of a stagnant job market that has decided it doesn’t need them. While AI is often touted as the excuse, it’s often just a cover for another simpler reason, like cutting costs. Whatever. The fact is, the belief that AI can — and will — displace white-collar jobs is a lie that’s been accepted by the masses, in part because of the huckster-like triangulations of scumbags like Sam Altman and Dario Amodei, and in part because credulous fuckwits in the media have repeated them verbatim. And I am concerned that the fear I’ve described is being exploited by companies like Replit and Cursor (which is also doing the exact same influencer marketing schtick, albeit not as aggressively as Replit), who are touting their services as a way for people to escape the precariousness of this current moment. First, the idea that we’re about to see the emergence of a thriving cottage industry of apps, all created and marketed by non-coders is fucking ludicrous. Forgive the causeness of my language, but I have to be blunt. Vibe-coded software is simply not good . Let’s suppose that someone deploys an app and there’s a critical security vulnerability that allows a threat actor to, say, exfiltrate all their customer information. How would they know? And if they became aware of it (presumably because said threat actor exploited said vulnerability), how would they fix it? Also, would the person who developed the app know that, under legislation like GDPR, they can be financially liable for data breaches ? Because they would be! And the whole point of the financial penalty system (at least, with respect to GDPR) is to be dissuasive — to act as a deterrent to other people who would be cavalier with other people’s data. I can very easily imagine a national data protection authority — like the UK’s ICO — giving someone a massive, massive fine in order to dissuade other people from deploying their own AI-generated, unvetted slop code. And then there’s the cost of actually building software. Replit’s business model allows customers to buy one of two subscriptions, each providing a certain number of credits. When you run out, you can either buy more credits, or simply rack up additional charges and pay at the end of the billing cycle. It’s entirely possible that someone will try to build their dream app, forget to cap their costs, and end up with a bill of hundreds or thousands of dollars. And no, that’s not an exaggeration . I do not see how this is any different than, say, someone buyin