The question everyone in media is asking is when, not if, they’ll be replaced by AI. This latent fear, as consistent and damaging as a tinnitus, is turning the dog-eat-dog world of news media into puma-eat-puma world, one that’s so amped up that hiring seems to be frozen and job applications go unanswered because of demand.
We’ve talked to a number of startups who might have an answer. Most recently I spoke with Duckbill AI, a service that offers household help. For example, it can look at your schedule and plan a packing list for your kids’ summer camp or help you figure out how to return something you don’t need. Basically it’s a concierge for daily life.
The trick is that the system uses AI and human copilots. The AI will generate a to-do list and make recommendations and the human double-checks the list for obvious hallucinations. This centaur-style work seems to achieve acceptable results for the company and they have a solid and relatively happy customer base.
This model is probably the default way we’ll be interacting with AI in the near future. Obviously making to-do lists isn’t the same as writing a 5,000-word article on politics, but the concept is the same. For example, you can use the AI to generate the basic text using your own reporting and then create something that looks and sounds like a news article in a few seconds. Then you can edit that same article by hand, turning the AI into a cub reporter.
But can you ever switch the situation? Can an AI replace you, the human in this equation and a cub reporter replace the AI?
Thus far the answer is “Probably not.” Consider this Donal Rumsfeld quote:
Reports that say that something hasn’t happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don’t know we don’t know. And if one looks throughout the history of our country and other free countries, it is the latter category that tends to be the difficult ones.
The difficultly inherent in our proposition is focused on the creative aspect of the potential project. But the AI doesn’t know what it doesn’t know in a way that is very Rumsfeldian and very dangerous because, in short, the AI doesn’t know anything. It knows what you want to read and make something that sounds operationally like a news story. But it will never be able to push back on a point or elaborate on a fact because it doesn’t know when to do this this, let alone why it should.
AI can generate a story and you can fix it. But you can’t generate a story and let the AI fix it. This important distinction might be what saves us all from extinction in the next few years, so keep your eyes peeled for situations when you’re being asked to exactly that — and make sure to push back.
Leave a Reply