Undergrad AI Predicting Physics Research Results

[ohwilleke]

TL;DR

Sabine Hossenfelder notes that AI can predict research results without reading papers. This is less impressive than it seems.

Sabine Hossenfelder's latest podcast is described as follows:​

Scientists Move Over: AIs Predict Research Results Without Doing Research

Scientific literature is growing rapidly, meaning scientists are increasingly unable to keep up with all of the latest developments in research. AI large language models, though, can read and “digest” information much more quickly than their human counterparts, making them the perfect tools to conduct massive literature reviews. Recent research shows they’re also very accurate at predicting the results of studies that they’ve never read before. Let’s take a look.

This sounds cool, until you think about it a bit more.

There are countless new HEP-EXP papers every week whose bottom line is: "No statistically significant deviation from the Standard Model was observed."

Likewise, there is no shortage of papers that say: "No statistically significant evidence of dark matter was observed and the cross-section exclusion for [the mass range mentioned in the article title] has been tightened to [one order of magnitude more than the last paper by the same research group]."

There are also loads of papers that conclude that: "The rare decay [from this particle to those particles as described in the title of this paper] was observed with more than five sigma significance."

These are, of course, merely the most elementary examples, but generally, the results of new physics research, while it needs to be done, doesn't have a lot of surprises.

Nobody writes a paper on sigma baryon decays that concludes: ". . . and then a large cat shaped hadron in a clown outfit walked out of the LHBb detector and ate a pineapple pizza," unless it is April 1st.

 

[Motore]

As far as I can see she is not explicitly talking about physics.