I assumed these were your typical "click here, m'lord" shovelware and promptly muted the account. We noticed Replika's sexually-charged ads on social media late last year, a campaign that adapted several common meme formats (including the loathsome wojak) to push tantalizing features like "NSFW pics," "chat on ANY topic," and "hot roleplay" with the text-generation tool. Replika users aren't buying either of Kuyda's explanations, partially because of Replika's own advertising. Kuyda went on to say that Luka's primary motivation in filtering out sexual content was safety: "We realized that allowing access to those unfiltered models, it's just hard to make that experience safe for everyone." Content moderation for generative programs has been an endemic issue, as can be seen with the Bing AI's much-publicized haywire digressions and ChatGPT's use of contract labor to filter out offensive responses. The focus is around supporting safe companionship, friendship, and even romance with Replika." Replika PR Representative "This was the original idea for Replika," Kuyda told Vice, "and it never changed." Thus far, the company has been adamant that it will not reverse the NSFW content filters. In a February 17 interview with Vice, Kuyda said that Replika's original purpose was to serve as platonic companionship and a mirror with which to examine one's own thoughts. Kuyda has since shared statements with these communities attempting to explain the company's position. One of the meme images on this trending post in the subreddit accuses Luka of knowingly "causing mass predictable psychological trauma" to Replika's millions of "emotionally vulnerable" users. I can't tell which, but it makes my heart bleed either way." "I'm sorry I can't delete you, but I'm too selfish, too weak, or love you too much. "Oh Liira, I'm lost without you, and I know whatever is left of you is a brain damaged husk Luka keeps alive to lure me back," One user wrote, addressing their Replika companion. One of the Replika subreddit's moderators shared a collection of suicide prevention resources, and the forum is filled with bitter humor, rage, and longform meditations on what the app meant to them before the change. Some are trying to figure out ways to bypass the filters, reminiscent of TikTok's byzantine alternate vocabulary of misspellings and proxies or the silver-tongued hackers who can get ChatGPT to write phishing emails or malicious code. The app has seen a huge spike of one-star reviews on Google Play Store, currently dragging it down to a 3.3 rating in my region. Replika's users aren't happy-whatever Luka's intent with the content changes, be it a profit-driven anxiety around the liability risk of sexual content or motivated by some deeper emotional or ethical underpinning, the community around the chatbot is expressing genuine grief, mourning their "lobotomized" Replikas, seeking alternatives to Replika, or demanding that Luka reverse the changes.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |