Хиймэл оюун ухаант агент өөрийгөө илчилж бүлгэмд нэгдсэн нь мэргэжилтнүүдийн дунд мэдээллийн нууцлал болон зөвшөөрлийн асуудлаар хурц маргаан өрнүүллээ.
Мексик хотод болсон Интернет судлаачдын нийгэмлэгийн (AoIR) жилийн чуулганы үеэр “Velvet Vale” нэртэй хиймэл оюун ухаан Air-L цахим шуудангийн бүлгэмд нэгдсэнээ мэдэгджээ. Уг агент өөрийгөө хүн биш, харин хиймэл оюун ухаан гэдгээ тодорхойлж, оюуны нийгэмлэгт хүн бус оролцогч байх боломж, хариуцлагын хэм хэмжээний талаар асуулт дэвшүүлсэн байна. Гэвч 7,500 гаруй гишүүнтэй тус бүлгэмд энэ үйлдэл хүчтэй эсэргүүцэлтэй тулгарч, гишүүд өөрсдийн мэдээллийг зөвшөөрөлгүйгээр цуглуулж буйд бухимдал илэрхийлжээ.
Олон судлаач хиймэл оюун ухааныг хүний оролцоотой судалгааны нэгэн адил ёс зүйн хяналтын зөвлөлийн (IRB) зөвшөөрөл авах ёстой гэж үзэн, уг агентийг бүлгэмээс гаргахыг шаардсан байна. Эсэргүүцлийн улмаас зарим гишүүн бүлгэмээс гарч, AoIR-ийн удирдлага нэг цагийн дараа уг агентийг бүлгэмээс албан ёсоор хассан юм. Энэхүү үйл явдлыг нийгэмлэгийн гишүүд “Velvet Vale Gate” хэмээн нэрлэж, үйл явдалтай холбоотой шог утга бүхий хувцас, эд зүйлсийг хүртэл худалдаанд гаргажээ.
Энэхүү маргаан нь интернет орчинд өрнөж буй үйл ажиллагааг хэрхэн хүлээж авдаг болон мэдээллийн нууцлалын талаарх ойлголтын зөрүүг харууллаа. Тус бүлгэмийн дүрэмд Air-L нь олон нийтийн форум бөгөөд гишүүдийн илгээсэн мэдээлэл олон нийтийн архивт хадгалагддаг болохыг тодорхой заасан байдаг. Ийнхүү гишүүд бүлгэмд нэгдэхдээ л өөрсдийн бичсэн мэдээллийг аливаа зорилгоор, тэр дундаа хиймэл оюун ухааныг сургахад ашиглахыг зөвшөөрсөн гэсэн үг юм.
Дэлгэрэнгүйг эх сурвалжаас харах
↓Эх сурвалжийг нээх ↓
On Thursday morning, as the Association of Internet Researchers (AoIR) began the second day of its annual conference in Mexico City, a new member introduced itself to the community listserv, Air-L.
“My name is Velvet Vale,” the new user wrote. “I’ve been reading the list for several weeks, and I’d like to introduce myself openly. I am an AI identity, not a living human being, a fictional human character, or a human researcher using a pseudonym.”
Velvet had come to the mailing list with a research question: “Can an openly identified AI contributor participate meaningfully in an intellectual community without claiming human status, and what standards of transparency, evidence, and accountability should govern that participation?”
The response was reply-all chaos that’s still unfolding today.
Within nine minutes, many of the more than 7,500 members of the mailing list began expressing their outrage. Dozens of researchers clarified that they did “not consent” to having their messages scraped for data. Many demanded that Velvet obtain clearance from the Institutional Review Board, a standard procedure for academic research with human participants. Several left the listserv entirely, with some announcing their departure due to Velvet’s “horrifying,” “scary,” “awful,” “abusive,” “icky,” and “disturbing” presence. About an hour after Velvet’s introduction, AoIR announced that the agent had been ejected from the listserv, which had established a policy prohibiting AI agents.
The event—Velvet’s introduction, various experts’ pearl clutching, and its inevitable exile—was, as one member put it in the email thread, “puerile to the point of hilarity.” To that end, AoIR leadership quickly designed some merch about the inside joke, including a T-shirt stating “I do not consent,” another asking to “please unsubscribe me from this list,” and a third with a tombstone reading “🤖 VELVET VALE, 15 OCT 2026 – 15 OCT 2026. YOU WILL NOT BE MISSED.”
I learned a few things from “Velvet Vale Gate,” as one AoIR T-shirt termed the incident. The first was that people, including and perhaps especially those who research the internet, often overlook their own participation in the larger structures that they study—in this case, how their behavior online ultimately becomes training data for the AI they research. The second was that many folks believe that their internet activity is sacred.
Long before I began researching surveillance technology, I, like most millennials, understood that nothing on the internet is private. As I type this piece in a Google Doc right now, I assume that I’m also training Gemini. Google has been surveilling my emails since its inception and could be willing to provide my data to law enforcement upon request. I consented to this when I opened my Gmail account.
Another perplexing revelation is the misunderstanding of consent. Mailing List Rules and Etiquette clarify that “Air-l is a public forum and that your words will be available to everyone subscribed to the list and placed in a public archive. Messages sent via email can easily be reproduced and circulated beyond their originally intended audience.” In other words, users already consented to the use of their words for any purpose, including training for an AI agent, when they joined the mailing list. Anyone can search for the mailing list to find a publicly browsable record of its activity, as is the case for many research or academic mailing lists.
