Стартап использует сгенерированного ИИ-персонажа для холодных писем, что вызывает критику за неэтичность

A couple weeks ago, my coworker Charlie got a cold email from a startup with a link to their “personalised” video pitch. I put scare quotes around “personalised” both because that’s literally how they described the video over email and because it turned out to be thoroughly impersonal trickery: the guy in the video was not a person at all. He was synthetic, either fabricated entirely or maybe a clone of the human who created him. The video is 90 seconds of this guy “Chase” scrolling around our landing page and pitching their product, a tool that generates launch videos for startups. “Hey Charlie, it’s Chase from Represent Studio. I was just on your website, and what you guys are building is super exciting.” Were we interested? Did Charlie want to hop on a quick call with “Chase”? The fake is impressive, even a little scary, but there are a few tells. Something about his glassy eyes darting around are not quite human. And the pitch felt a little too pointedly casual. Also, “We’ve had some of the biggest launches in the Silicon Valley.” The Silicon Valley? I had recently binge-listened to Shell Game, Evan Ratliff’s podcast in which he documents his experimental startup run entirely by AI employees, so I was primed to sniff out just this sort of synthetic human fakery. Charlie noticed it, too. You can watch the video yourself—I’m sure “Chase” will appreciate the publicity. To satisfy our curiosity, Charlie replied: “Hi. Was this an AI video?” No prodding needed, we got our answer: “Yes, this is a video generated by an AI system we built internally. Curious, did it resonate with you?” Well, “Chase,” it certainly got our attention! Pretending to be human is unethical Good thing we have Robin Sloan to call a spade a spade. From his pop-up newsletter last month: It’s plainly unethical to use robots to perform simulated human marketing outreach. I know people were doing that long before the arrival of LLMs; it was unethical then, too. If you want to use a robot to send me a customized pitch, write: “Hello, this is an automated message from Robot Corporation. We think you’re a potential customer, so we’d like to give you some information”—and so on. In my estimation, that approach is still annoying, but ethically okay. By now, AI slop is a well-circulated idea. It’s helpful to have a name for it so that we can call it out—see, for example, Stop Sloppypasta. Slop is a real problem, in part because we’re now having to constantly question the source and truth of whatever we read (and increasingly, see and hear through screens and speakers). It’s a tax on our daily allotment of mental energy. As Jane Friedman writes, we need relief from this constant epistemic vigilance. Sloan warns us about a “scenario in which AI systems pollute human communication channels to the degree that they’re no longer reliable or even usable.” He calls it “a profound digital-ecological crisis.” It is an internet crisis, one that majorly bums me out, but there’s at least one thing we can do. Stop pretending! Going to market ethically This year I’ve observed Steve and Charlie selling our product, effortfully and ethically. They talk to lots of customers, sometimes to sell and other times for product feedback. When you hear Steve and Charlie talk about Val Town you can tell they themselves actually use and love the product, which of course makes it easier to sell earnestly. Almost four years in, Val Town can be called a startup success insofar as we have paying customers, who work at companies we respect, building neat things. We’re also proud of the product and like using it ourselves. But it’s not the kind of inevitable startup success where I’m already penciling in a down payment on a mortgage from an IPO in a few years. We’re working on product-market fit, as Steve writes about in his monthly public investor updates, and we think go-to-market engineering might be that fit. Which is why it feels especially timely now to call out our distaste for dishonest slop. And to be clear, using AI to build go-to-market tools is not what we take issue with. We have templates that do just that, like to qualify leads or enrich signups. It’s the pretending that’s bad. When our go-to-market motion has started to feel too automated and disingenuous, we’ve pulled back. We want to find people for whom Val Town is the right tool, but we don’t want to waste their time or deceive. I’m sorry to make an example of “Chase,” and I don’t think he’s a bad person for trying to use new tech to grow his startup. It’s really hard to get people to try your thing. I’m going to keep quoting Sloan here, because he’s just so spot on. He writes that sending customized, AI-generated cold email “is subtler than a one-size-fits-all promo blast, but it sucks way worse, because it’s fundamentally dishonest.” And these offenders like “Chase” are ruining it for everyone: Besides being plainly rude and dishonest, these messages “pee in the pool” of internet communication, making it more difficult for sincere creators to send authentic emails about their projects, simply by raising the “noise floor” of simulation and bullshit. Cold emails are totally fine — either make them sincerely personal or sincerely impersonal. Nobody wants to hear from your AI bot, least of all when it’s pretending to be you, laying it on thick. So please, stop pretending. We're hiring! Are you an infra engineer who cares about the joy of programming?