三星公布“人造人”項目:類似真人,可以表達情感
三星全新的神秘項目Neon終于揭開面紗。
據(jù)悉,這種人形AI聊天機器人由三星技術(shù)和高級研究實驗室(又名STAR Labs)開發(fā)。
該公司將其Neon技術(shù)描述為“通過計算創(chuàng)建的虛擬人,其外觀和行為類似于真實的人類,并具有表達情感和智慧的能力?!?/p>
“Neon?is like a new kind of life,” says STAR Labs CEO Pranav Mistry in the release. “There are millions of species on our planet and we hope to add one more.”
“Neon就像是一種新的生命,” STAR Labs首席執(zhí)行官普拉納夫?米思特里在發(fā)布中說道?!拔覀冃乔蛏嫌袛?shù)以百萬計的物種,我們希望再增加一個。”
Each Neon avatar is “computationally generated” and will hold conversations with users while displaying “emotions and intelligence,” says the company. Their likenesses are modeled after real humans, but have newly generated “expressions, dialogs, and emotion.” Each avatar (known individually as “NEONs”) can be customized for different tasks, and is able to respond to queries “with latency of less than a few milliseconds.” They’re not intended to be just visual skins for AI assistants, but put to more varying uses instead:
該公司表示,每個Neon分身都是“計算生成的”,將與用戶進行對話,同時顯示“情感和智慧”。Neon分身的相似之處是按照真實的人類建模的,但是具有新生成的“表情,對話和情
感”。每個分身(分別都被稱為“ NEON”)都可以針對不同的任務(wù)進行定制,并且能夠在幾毫秒內(nèi)“以更少的延遲響應(yīng)查詢”。它們不只是有了視覺外觀的AI助手,而是被用于更多用途之中:
“In the near future, one will be able to license or subscribe to a NEON as a service representative, a financial advisor, a healthcare provider, or a concierge. Over time, NEONs will work as TV anchors, spokespeople, or movie actors; or they can simply be companions and friends.”
“在不久的將來,人們將能夠授權(quán)或訂閱Neon作為服務(wù)代表、財務(wù)顧問、醫(yī)療保健提供者或禮賓。隨著時間的推移,Neon將充當(dāng)電視節(jié)目主持人、代言人或電影演員;或者他們可以只是人類的伙伴和朋友。”
In its PR, the company veers into strange territory is in its description of the avatars’ underlying technology. It says it’s using proprietary software called “Core R3” to create the avatars, and that its approach is “fundamentally different from deepfake or other facial reanimation techniques.” But it doesn’t say how the software does work, and instead relies on wishy-washy assurances that Core R3 “creates new realities.” We’d much rather know if the company is using, say, high-resolution video captures pinned onto 3D models or AI to generate facial movements — whatever the case may be.
在公關(guān)中,該公司轉(zhuǎn)向了一個陌生領(lǐng)域——在其描述中的分身基礎(chǔ)技術(shù)。該公司稱其正在使用名為“Core R3”的專有軟件來創(chuàng)建分身,并且其方法“與Deepfake或其他面部模擬技術(shù)有根本不同?!钡玈TAR Lab并未說明該軟件如何工作,而是依賴于空泛地保證Core R3“創(chuàng)造了新的現(xiàn)實”。我們更想知道該公司是否在3D模型或人工智能上使用所謂的高分辨率視頻捕獲技術(shù)來生成面部表情——無論實例是什么。
多知道一點
普拉納夫?米思特里
普拉納夫?米思特里?Pranav Mistry,他 2009 年因兩個名為“第六感科技”(The Sixth Sense Technology)的 TED 演講在中國技術(shù)領(lǐng)域被廣泛關(guān)注,他個人也是三星集團的全球高級副總裁。
普拉納夫?米思特里 Pranav Mistry 是一位印度裔的計算機科學(xué)家和發(fā)明家,現(xiàn)任三星旗下的獨立研發(fā)機構(gòu)STAR Labs的創(chuàng)始CEO兼總裁,他也曾擔(dān)任三星電子的全球高級副總裁。Pranav 畢業(yè)于印度理工學(xué)院(IIT)和麻省理工學(xué)院(MIT),是日本科學(xué)技術(shù)廳客座研究員,曾獲得諸如世界經(jīng)濟論壇的“世界青年領(lǐng)袖”,“全球電訊行業(yè)50人”等嘉獎,被《大眾科學(xué)》、《麻省理工科技評論》等國際媒體譽為當(dāng)今最重要的創(chuàng)新家之一。