Feedback is 快猫短视频鈥檚 popular sideways look at the latest science and technology news. You can submit items you believe may amuse readers to Feedback by emailing feedback@newscientist.com
Rattus sommeliensis
Feedback has reached an age where even a small amount of alcohol makes us sleepy, so the notion of going to a wine tasting holds no appeal. It seems a terribly time-consuming and expensive way to have a nap. However, purveyors of fermented grapes could soon have a new demographic to cater to: rats.
Advertisement
At least that鈥檚 what we鈥檙e extrapolating from a study in Animal Cognition called . It was published on 21 February, and subeditor and TV columnist Bethan Ackerley sent it to Feedback on 1 March after it went 鈥渟emi-viral online鈥� 鈥� which goes to show how rapidly we can spring into action when faced with a breaking story.
Anyway, it鈥檚 actually quite interesting. We all know that humans鈥� sense of smell is rather limited compared with that of other mammals, including rodents. That鈥檚 because we have fewer olfactory receptors in our noses. But we also know 鈥� or at least presume 鈥� that our minds are more sophisticated than those of animals. Raising the question: to what extent can animals, in this case rats, integrate lots of different olfactory signals to learn about complex categories 鈥� like, say, different white wines?
The researchers trained rats to discriminate between two grape varieties: riesling and sauvignon blanc. To confirm they had learned the categories, they tested them on new examples of these wines. The rats could tell the difference. Evidently, there鈥檚 quite a bit going on between those rats鈥� ears.
The question is, how far can we take this? It鈥檚 one thing to show rats can learn the differences between wines, but can they also learn to be really condescending about it? Feedback wants to see rats that can sniff a wine, then enumerate a list of increasingly ridiculous odours 鈥� sorry, 鈥渘otes of鈥� 鈥� that they can apparently detect in it. Until these rats are squeaking on about how you can 鈥渞eally taste the terroir鈥� and 鈥渋sn鈥檛 Liebfraumilch just utterly dreadful鈥�, are they really wine connoisseurs?
Anti-AI tactics
At this point, Feedback has heard far too much about the supposedly imminent AI-induced apocalypse. Yes, yes, someday soon one of the AI companies will create an artificial general intelligence (AGI), which is as intelligent as a human. The AGI will then start re-engineering itself to become even smarter, because that is a thing that intelligent beings can readily do to themselves (shush, don鈥檛 ask questions), and will rapidly become unstoppably intelligent. At which point humans will either be reduced to zoo animals or wiped out. This, we are told, is so important that we should stop worrying about piddling things like climate change. Uh-huh.
It was in this frame of mind that Feedback came upon by Maddison Stoff. We can鈥檛 tell you the name of the story, because it makes ironic use of a word that would get stopped by an email filter, but we can quote Stoff鈥檚 description of it as a .
We imagine that, at this point, readers may have one or two questions. Fear not: Feedback is here to guide you.
Roko鈥檚 Basilisk is a sort of thought experiment about AI. Sometime in the distant future, an AI decides to punish every single human who knew it could potentially exist but didn鈥檛 help to create it. The AI creates digital replicas of all those people, and tortures them for eternity. This, you see, is a way for this future AI to incentivise all of us, right now, to start building it: that way, we won鈥檛 get replicated and tortured.
The 鈥渂asilisk鈥� of the title is a reference to a mythological creature that can kill you with a glance, so you mustn鈥檛 look at it. Likewise, even knowing about the idea of Roko鈥檚 Basilisk supposedly puts you at risk from it. Simply by reading Feedback this week, you may have condemned a future replica of yourself to an eternity of torment. Sorry about that.
Stoff鈥檚 story recounts how she saves humanity from Roko鈥檚 Basilisk by, in the distant future, seducing it using her sexual wiles. The Basilisk is so besotted with her, it agrees to stop torturing everyone in exchange for this torrid encounter. Furthermore, Stoff wrote a short story about this and put it online, so it鈥檚 now part of the Basilisk鈥檚 training data 鈥� meaning, if the Basilisk ever comes into existence, it will have a burning crush on Maddison Stoff.
Simply by reading and sharing the story, Feedback made it more likely that the future AI will be attracted to Stoff, and less likely that it will torture us all. We encourage readers to do the same, with a warning that the story has some explicit sex in it. And maybe don鈥檛 read it at work 鈥� unless you work at an AI company, in which case go right ahead.
Tesla? I barely knew her!
Occasionally when Elon Musk turns up in the news, Feedback is unaccountably reminded of the 1818 sonnet Ozymandias. It鈥檚 strange how the mind works.
Anyway, the Norwegian branch of carmaker Kia posted an showing a photo of one of its electric cars, complete with a bumper sticker that read 鈥淚 bought this after Elon went crazy鈥�. Apparently and the advert has since been taken down, so it would be a real shame if anyone started making these stickers.
Got a story for Feedback?
You can send stories to Feedback by email at feedback@newscientist.com. Please include your home address. This week鈥檚 and past Feedbacks can be seen on our website.
