- This topic has 2 replies, 2 voices, and was last updated 23 hours, 52 minutes ago by
.
-
Posts
-
October 2023 – First supernova detected, confirmed, classified and shared by AI “New artificial intelligence tool removes humans from entire search, discovery process”
https://ciera.northwestern.edu/2023/10/16/first-supernova-detected-confirmed-classified-and-shared-by-ai/January 2025 – amateur astronomer points out AI discovered “transient” AT 2024agoh is actually a high proper motion star
https://www.wis-tns.org/object/2024agohThat’s not quite fair on the robots!
The first link refers to SN 2023tyk which went all the way to spectroscopic confirmation without human intervention:
https://www.wis-tns.org/astronotes/astronote/2023-265
The second link is to AT 2024agoh which is a transient candidate with no spectroscopic confirmation. Certainly whatever AI/ML algorithms they are using needs to be a bit smarter and check more catalogues.
Give it time and AI will take over the world. Our current government certainly thinks it is our saviour.
In the meantime these systems are being used without oversight or basic checks that any observer should be doing. (The TNS system has no gatekeeper and relies on the accuracy of the submitted data). The issues with this observation are multiple and obvious without needing any further data to anyone with basic knowledge of how transients behave (A mag 14 object which suddenly appears with no sign of a precursor or host galaxy and then stays at identically the same brightness for several weeks should already have raised suspicions.) The ghost negative “missing star” in the (observed – archive) difference image and the double colour star images in the DSS 2 colour archive image of the region made it obvious was going on (even before taking into account the proper motion from Gaia and the 2MASS image showing it in yet another position) yet the operators of this system “followed” it for 4 weeks before finally triggering it as “discovery”.
- You must be logged in to reply to this topic.