On 9/9/26 9:35 PM, Solar Designer wrote: > Hi, > > I've just reluctantly approved 3 AI slop postings by Eve > <[email protected]>, who is an "automated security researcher". I don't > know if there's any value in those, but I cannot rule that out. The > value could end up historical preservation of what these reports looked > like at the dawn of AI security research. > > These messages lack a Date header, but they arrived to the list server > on September 9 starting 6am UTC and until 11am UTC. For two of these, > she sent several versions of the same posting. I diff'ed them and ended > up approving the latest versions, which include more content, but also > have some assorted wording differences and some removals. > > So where she writes she notified a vendor on September 9 and received no > response, it means she gave the vendor no time to respond or didn't > actually notify them. Many AIs still fail at (or don't try) keeping > track of time and of what they did before or are going to do later.
Leaving aside the colorfully professional joke email domain used, I cannot help but feel it would be beneficial to avoid anthropomorphizing the computer program (while simultaneously complaining they are either incoherent or simply probably wrong). Volunteering for free to treat bots like humans feels like the opposite of what you seem to be saying you want... -- Eli Schwartz
OpenPGP_signature.asc
Description: OpenPGP digital signature
