The UCLA team, Taylor Webb, Keith Holyoak, and Hongjing Lu, relied on a large collection of ways that past research has tested humans’ ability to reason via analogy. The classic form of this is the completion of a comparison—think “cold is to ice as hot is to ____”—where you have to select the best completion from a set of options.
Related tests involve figuring out the rules behind transformations of a series of letters. So, for example, if the series a b c d is transformed to a b c e, then the rule is to replace the last letter of the series with its alphabetical successor. The participant’s understanding of the rule is tested by asking them to use the rule to transform a different set of letters. Similar tests with numbers can involve complex rules, such as “only even numbers in order, but can be ascending or descending.”
On all of these tests, GPT-3 consistently outperformed undergrads, although the margins varied depending on the specific test involved. The researchers also found that the software could develop rules based on a series of numbers, and then apply them to a different domain, such as descriptions of temperatures like “warm” and “chilly.” They conclude that “these results suggest that GPT-3 has developed an abstract notion of successorship that can be flexibly generalized between different domains.”
Many of the fundamental features of life don’t necessarily have to be the way they are. Chance plays a major role in evolution, and there are always alternate paths that were never explored, simply because whatever evolved previously happened to be good enough. One instance of this idea is the genetic code, which converts the information carried by our DNA into the specific sequence of amino acids that form proteins. There are scores of potential amino acids, many of which can form spontaneously, but most life uses a genetic code that relies on just 20 of them.
Over the past couple of decades, scientists have shown that it doesn’t have to be that way. If you supply bacteria with the right enzyme and an alternative amino acid, they can use it. But bacteria won’t use the enzyme and amino acid very efficiently, as all the existing genetic code slots are already in use.
In a new work, researchers have managed to edit bacteria’s genetic code to free up a few new slots. They then filled those slots with unnatural amino acids, allowing the bacteria to produce proteins that would never be found in nature. One side effect of the reprogramming? No viruses could replicate in the modified bacteria.