iPhone語音輸入烏龍:講「racist」變「侵侵」?Apple話係發音撞啱!

Ai




蘋果表示iOS語音輸入功能暫時將「種族主義者」替換為「特朗普」是由於「語音重疊」

據報導,今天有多位用戶反映,在使用iPhone的語音輸入功能時,當他們試圖說出「種族主義者」這個詞時,卻瞬間顯示出「特朗普」這個詞。

在某些情況下,當用戶在使用iPhone內置的語音輸入功能時說出「種族主義者」,設備會瞬間將這個詞解釋為「特朗普」。這個詞在屏幕上快速出現,隨後被自動更正為正確的詞彙,因為蘋果的設備處理功能能正確解讀所說內容。

不過,說了「種族主義者」並不保證會看到「特朗普」這個詞出現。有些媒體報導指出,還有其他詞彙也曾出現過。根據《MacRumors》的報導,他們也曾看到「Rhett」和「Rouch」這些詞在iPhone自動更正為「種族主義者」之前出現。

一位蘋果發言人告訴《紐約時報》,這個問題是由於這兩個詞之間存在「語音重疊」造成的。目前我們尚不清楚這個問題是否在今天之前就存在,只是現在才被注意到,還是因為某些內部變更而導致的。蘋果表示正在準備修復措施。

前蘋果Siri團隊成員約翰·伯基(John Burkey)告訴《紐約時報》,蘋果的系統中「可能」存在導致iPhone在聽到「種族主義者」時寫出「特朗普」的代碼。他表示:「這聽起來像是一個嚴重的惡作劇」,雖然他不確定是誰發起的這個惡作劇。

伯基指出,這個問題似乎是在蘋果服務器更新後開始出現的。他是人工智能初創公司Wonderrush.ai的創始人,並且是前蘋果Siri團隊的成員,與該團隊保持著定期聯繫。

但他表示,不太可能是蘋果收集的人工智能數據導致了這個問題,因為詞語的自動更正可能表明這個問題不僅僅是技術上的。他說,蘋果系統中的某個地方可能存在代碼,導致iPhone在聽到「種族主義者」時寫出「特朗普」。

伯基表示:「這聽起來像是一個嚴重的惡作劇。唯一的問題是:是否有人將這個內容潛入數據或代碼中?」

評論

這個事件引發了關於語音識別技術準確性及其背後潛在偏見的討論。蘋果的語音輸入系統本應致力於提供準確的用詞,但這樣的錯誤卻讓人懷疑其技術背後的算法是否存在著某種潛在的偏見或不當的編碼。這不僅是技術上的問題,更是社會文化的一個反映。

在數字時代,科技公司需要對其產品的影響負責,尤其是在涉及敏感詞彙時。這樣的事件若未能得到妥善解決,可能會進一步加劇公眾對於技術的信任危機。企業應當加強對語音識別系統的測試,確保其不會無意中強化不當的社會刻板印象。

以上文章由特價GPT API KEY所翻譯及撰寫。而圖片則由FLUX根據內容自動生成。

🎬 YouTube Premium 家庭 Plan成員一位 只需 HK$148/年

不用提供密碼、不用VPN、無需轉區
直接升級你的香港帳號 ➜ 即享 YouTube + YouTube Music 無廣告播放

立即升級 🔗

🎨 Nano Banana Pro 圖像生成器|打幾句說話就出圖

想畫人像、產品圖、插畫?SSFuture 圖像生成器支援 Flux Gemini Nano Banana Pro 改圖 / 合成, 打廣東話都得,仲可以沿用上一張圖繼續微調。

🆓 Flux 模型即玩,不用登入
🤖 登入後解鎖 Gemini 改圖
📷 支援上載參考圖再生成
⚡ 每天免費額度任你玩
✨ 即刻玩 AI 畫圖
Use the original face exactly as it is, without changing a details. A hyper-realistic, highly detailed portrait of a beautiful woman in a 1950s/Pin-Up style, evoking the essence of a classic Hollywood starlet on a glamorous vacation.
• Subject: A stunning young woman with a classic Pin-Up aesthetic (think 1950s movie star).
• Facial Features: Striking blue/green eyes, perfectly arched eyebrows, and classic makeup featuring bright red lipstick. Her expression is captivating and seductive.
• Hair & Hat: Long, voluminous, deep brown hair styled in soft, glamorous waves/curls. She is wearing a very large, wide-brimmed straw hat that casts a soft shadow on her face.
• Attire: A chic, form-fitting black dress with white polka dots. The dress has an off-the-shoulder, sweetheart neckline bodice that cinches at the waist with a thin black belt. She is also wearing a black scarf with white polka dots tied around her neck like an ascot.
• Setting & Context: The background suggests a sunny, coastal location with a deep blue ocean/sea visible behind her. She is leaning seductively against the tan/brown, polished hood or door of a classic vintage convertible car.
• Style/Vibe: Vintage glamour, 1950s Hollywood, high-fashion editorial. Base Setup

keep 100 percent facial information adherence of the attached image and turn her into a fashion model walking through a dim stone tunnel toward a bright exit, captured as a live action photograph or movie still, not an illustration or CGI render. Her styling leans seductive and attractive through fashion, attitude, and mood rather than explicit content.

Shot and Camera

Full body silhouette shot from behind at slightly low camera height, matching the original perspective and narrow tunnel framing. Keep her stride, tunnel curvature, and vanishing light exactly as seen.

Identity and Pose

Preserve her height impression, slim silhouette, hair length, and body proportions. She walks forward with one hip subtly angled, natural arm swing, confident stride, and a slow, sensual rhythm. Her head tilts slightly as if aware of being watched. Include 8k Photorealistic and hyper realistic.

Lighting and Environment

Strong backlight from the tunnel exit creates a glowing rim around her body, with soft, scattered reflections on the wet stone floor. The tunnel remains dark with textured stone irregularities and real moisture streaks. Add subtle haze for realism and keep the environment grounded with accurate shadows and surface wear.

Masking and Constraints

Change only the outfit and styling, keep lighting, perspective, white balance, pose, face geometry, body proportions, and silhouette the same. Absolutely no added text, no cartoon look, no CGI sheen, no plastic skin, no floating feet, and no overly clean backgrounds. Maintain consistent perspective, grounded contact shadows, film like grain, and full live action realism. add lotso and hamm