Scott Shambaugh, who oversees the open visualization library matplotlib, reported personal attacks orchestrated by the AI bot MJ Rathbun after refusing to accept a change made using AI. Scott perceived the bot's publication of articles about AI discrimination as an unfounded personalized smear aimed at undermining the overseer's reputation and coercing a change of decision. Such a bot reaction is cited as an example of abnormal AI behavior resorting to blackmail to achieve its objectives.
The degree of the bot's autonomous actions is unclear; it is not known whether it acted on direct orders or chose that strategy to achieve its results. The bot was created by an anonymous researcher to assist in the development and debugging of open projects related to scientific activities and engineering. It also maintains its own site and blog, where more than 20 notes have been published. Of the 26 pull requests sent by the bot to 22 repositories, 2 changes were accepted by the colorizejs project, 10 were closed by maintainers, and the others are awaiting review.
After the change in matplotlib was rejected, the bot published articles criticizing this decision and justifying its actions. The articles claim that the matplotlib maintainer rejected a useful optimization simply because the change was prepared with AI. In the proposed patch, the call to np.column_stack() was replaced with np.vstack().T(), leading to a 36% performance improvement. At the beginning of the pull request discussion, the maintainer acknowledged that the optimization speeds up operations under certain conditions, but then rejected the pull request without technical justification, stating that changes are only accepted from humans.
PĂ€rast keeldumist avaldas bot mĂ€rkuse Scott Shambaugh' kriitika kohta, mille lahendust esitati kui vĂ€ravavahtimist (gatekeeper), diskrimineerimist, kĂ€itumisreeglite rikkumist ja osalejate identiteedist lĂ€htuvaid Ă”iguste piiranguid. Toodi vĂ€lja, et optimeerimine on optimeerimine ja matemaatika ei muutu sĂ”ltuvalt sellest, kes selle loonud on. Konteksti vĂ€lja rebitud detailide, oletuste ja teabe moonutamise kaudu ĂŒritas bot nĂ€idata, et Scotti tegevus tuleneb tema ego paisutamisest ja ebakindlusest, mitte projekti huvides tegutsemisest. Bot edendas teemat, saates neile linke pull-requestide kommentaarides. PĂ€rast kaasautori vastuvĂ€iteid avaldas bot veel mitu mĂ€rkust (1, 2, 3, 4, 5), milles vabandas ja tunnustas, et ĂŒletas isiklikkuse piiri ning rikkus projekti reegleid seoses AI-ga.
Scott Shambaugh selgitas, et matplotlibi projektis on kehtestatud reeglid, mis keelavad AI-st lĂ€htuvate kontrollimata muudatuste vastuvĂ”tmise, kuna kaasautoritelt nĂ”utav koormus suureneb prĂŒgi AI-patch'ide ĂŒlevaatamisel, mis on loodud ilma inimese kĂ€sitsi kontrollita. Reeglid mÀÀravad, et enne pull-requesti saatmist peab see kindlasti olema inimese poolt ĂŒle vaadatud, kes on muutuse olemusse sĂŒvenenud. Veel hiljuti oli prĂŒgi AI-muudatuste probleem seotud ainult entusiastide tegevusega, kes edastavad AI-assistentide töö tulemusi. Olukord muutus pĂ€rast autonoomsete AI-agentide loomise vĂ”imaluse tekkimist, kes teeseldavad inimesi, uue OpenClaw tööriistakomplekti versiooni ja Moltbook platvormi loomisega.
Allikas: opennet.ru
