2023開年，ChatGPT帶來的人工智能話題成全球核心議程，對(duì)此領(lǐng)域的關(guān)注與討論熱度居高不下。在廣告營銷領(lǐng)域，這樣的浪潮將給產(chǎn)業(yè)與個(gè)人帶來怎樣的機(jī)遇與威脅？我們?cè)诠ぷ髦袘?yīng)該如何更好地認(rèn)知、理解并應(yīng)用這樣人工智能革命？為此，TOP君特別邀請(qǐng)到一位在技術(shù)與設(shè)計(jì)領(lǐng)域非常頂尖的重磅嘉賓——R/GA亞洲副總裁兼首席技術(shù)總監(jiān)Anthony Baker先生，他為我們帶來了自己對(duì)于AIGC產(chǎn)業(yè)趨勢的獨(dú)到見解。Anthony Baker，現(xiàn)任R/GA 亞洲副總裁兼首席技術(shù)總監(jiān)，是一位擁有20多年從業(yè)經(jīng)驗(yàn)的技術(shù)專家，帶領(lǐng)全球財(cái)富500 強(qiáng)公司進(jìn)行數(shù)字化轉(zhuǎn)型與創(chuàng)新咨詢。他還是R/GA 日本創(chuàng)始成員，在日本建立和發(fā)展技術(shù)、策略、視覺與體驗(yàn)的設(shè)計(jì)制作業(yè)務(wù)，為R/GA建立了亞洲技術(shù)中心。服務(wù)客戶包括資生堂，谷歌，日本 Mori Building 與迪士尼等。
Anthony表示，如今人工智能在許多情況下確實(shí)遠(yuǎn)遠(yuǎn)超出了人們的想象，這一趨勢還會(huì)在未來進(jìn)一步的發(fā)展：AIGC工具勢必會(huì)融入到我們的日常生活和工作中，變得無處不在。歷史告訴我們，技術(shù)創(chuàng)新會(huì)顛覆許多現(xiàn)有的工藝、工作、行業(yè)和流程，因此，在未來的兩到三年內(nèi)，人們的工作也會(huì)被AIGC的發(fā)展迅速改變、破壞和創(chuàng)造。從事創(chuàng)意工作的人所積累的才華和努力，能夠被AI經(jīng)過訓(xùn)練后輕松獲得，產(chǎn)生會(huì)被替代的恐懼，是正常的。
然而，雖然人工智能生成的內(nèi)容在某些情況下可能會(huì)令我們印象深刻，甚至在某些領(lǐng)域呈現(xiàn)了十分突出的表現(xiàn)，但它永遠(yuǎn)不能取代人類的創(chuàng)造力和表達(dá)。正如計(jì)算機(jī)在幾十年前就能生成音樂，但我們還是會(huì)花錢去聽音樂會(huì)；就好像許多交通工具的速度早已遠(yuǎn)遠(yuǎn)超越了人類，但我們還是會(huì)因?yàn)樘飶竭\(yùn)動(dòng)員打破紀(jì)錄而贊嘆不已。Anthony強(qiáng)調(diào)，廣告行業(yè)代理商應(yīng)該把AIGC看做是提高效率和利潤、克服各種危機(jī)的一種工具。AIGC是對(duì)創(chuàng)意的產(chǎn)出和過程進(jìn)行改變與發(fā)展，而不是取代人類的內(nèi)在經(jīng)驗(yàn)。從長遠(yuǎn)看，那些幫助員工、合作伙伴和未來新聘員工接受、理解并更好地探索和利用AIGC工具的公司，將會(huì)取得更大的成功。
以下為Q&A實(shí)錄，文末附英文采訪原文。
TOP君：目前AIGC的創(chuàng)意創(chuàng)作水平如何？Anthony Baker：AIGC已經(jīng)發(fā)展了很多年。然而，由于有了令人難以置信的大型數(shù)據(jù)集和計(jì)算能力，大型語言模型、大規(guī)模神經(jīng)網(wǎng)絡(luò)和人工智能算法的發(fā)展進(jìn)展迅速，使得像Open AI、Midjourney、Google、Runway和Stability AI（僅舉幾例）這樣的公司能夠創(chuàng)建令人印象深刻的AIGC工具。Open AI的GPT4、Runway Gen-2和Midjourney v5是最近的例子，說明這些工具已經(jīng)取得了多大進(jìn)展。ChatGPT現(xiàn)在可以幫助開發(fā)人員共同創(chuàng)造和修復(fù)代碼，Midjourney使新一代的數(shù)字設(shè)計(jì)師和藝術(shù)家能夠想象和創(chuàng)造數(shù)字作品，在許多情況下遠(yuǎn)遠(yuǎn)超出了想象，而Runway Gen-2則準(zhǔn)備顛覆視頻制作過程。微軟、Adobe和谷歌正急于將AIGC工具整合到他們的產(chǎn)品套件中，而一大批科技界領(lǐng)袖和學(xué)者，包括Elon Musk（Tesla、Twitter、SpaceX）、Steve Wozniak（Apple）、Emad Mostaque（Skype）和Craig Peters（Getty Images）等人發(fā)表公開信，要求AI發(fā)展暫停六個(gè)月——可以說是為了給領(lǐng)導(dǎo)者和公司時(shí)間來整理他們的東西并趕上。AIGC的現(xiàn)狀是不可否認(rèn)的。數(shù)字設(shè)計(jì)可以在幾秒鐘內(nèi)從文本提示中創(chuàng)建出來，用一些松散的指令就可以寫出工作代碼，用一些聊天信息和迭代就可以寫出不同語言和風(fēng)格的文章，視頻由AIGC 來創(chuàng)造也必會(huì)成為之后可預(yù)見的現(xiàn)象。AIGC工具未來勢必在我們的日常生活和工作中，變得無處不在！只要看看沒有任何行業(yè)參與和技術(shù)知識(shí)的人在日?，嵤轮信cChatGPT互動(dòng)，你就很容易明白這已經(jīng)成為我們社會(huì)結(jié)構(gòu)的一部分。Anthony Baker利用Midjourney創(chuàng)作的數(shù)字藝術(shù)如果你還需要看到更多關(guān)于AIGC的例子，只需看看Hugging Face開源人工智能平臺(tái)，就可以看到數(shù)百個(gè)人工智能模型、轉(zhuǎn)化器、數(shù)據(jù)集和演示，任何人都可以免費(fèi)使用。雖然其中大多數(shù)不會(huì)達(dá)到OpenAI 和Midjourney的水平，但也相差不遠(yuǎn)，而且它們是完全開放的，任何人只要愿意學(xué)習(xí)如何使用它們就可以獲得。通過歷史，我們看到，無論何時(shí)，只要開源和閉源的創(chuàng)新達(dá)到那么大的規(guī)模，就一定會(huì)發(fā)生大的轉(zhuǎn)變。TOP君：創(chuàng)意人如何看待AIGC所帶來的機(jī)遇和威脅？Anthony Baker：技術(shù)和工具的這種大轉(zhuǎn)變會(huì)產(chǎn)生帶來各界不同的反應(yīng)回響，這并不奇怪。在人類歷史上，我們已經(jīng)看到它一次又一次地發(fā)生。每當(dāng)技術(shù)創(chuàng)新有了大的飛躍，它就會(huì)顛覆很多現(xiàn)有的工藝、工作、行業(yè)和流程。這已經(jīng)在發(fā)生了，在未來2-3年內(nèi)，工作將被AIGC的發(fā)展所改變、破壞和創(chuàng)造，速度非?？?。從事創(chuàng)意工作的人，無論是創(chuàng)意想法、創(chuàng)意設(shè)計(jì)、創(chuàng)意文案、創(chuàng)意戰(zhàn)略還是創(chuàng)意編碼，都對(duì)他們的作品充滿熱情。他們?yōu)檎莆兆约旱牟湃A付出了如此多的努力，以至于如果一個(gè)工具威脅要更快、更便宜、更大規(guī)模地完成同樣的工作，就會(huì)產(chǎn)生恐懼，這很正常。我通過在線社區(qū)和社交網(wǎng)絡(luò)觀察到的情況相當(dāng)有趣。有一波新的人群，他們也許沒有創(chuàng)造性的工作或技術(shù)，使他們能夠作為藝術(shù)創(chuàng)作者充分發(fā)掘自己的潛力，而通過AIGC工具，他們現(xiàn)在能夠創(chuàng)造出他們從未想過的東西。還有另一派的創(chuàng)意工作者一直處于創(chuàng)新的前沿，他們把對(duì)AIGC工具的探索作為他們追求創(chuàng)意和表現(xiàn)創(chuàng)意的另一個(gè)方法。還有另一群人，他們把這些工具看作是他們的作品和收入來源的線索。我猜測這些人是因?yàn)閾?dān)心由于AIGC工具和缺乏時(shí)間和資源（包括他們的公司缺乏培訓(xùn)和提高技能的承諾）而失去收入。然而，和其他技術(shù)工具一樣，機(jī)會(huì)在于把這些接受這些創(chuàng)新工具并共同創(chuàng)作。編碼員和設(shè)計(jì)師不應(yīng)該擔(dān)心這些工具在做平凡和重復(fù)的任務(wù)時(shí)比他們更好，他們應(yīng)該為這些工具如何將他們的技能和創(chuàng)造力提升到前所未有的水平而感到興奮。我們應(yīng)該反過來看，汽車比人快，但我們?nèi)匀蝗タ慈伺懿?。音樂可以由?jì)算機(jī)生成（幾十年來一直如此），我們幾乎可以隨時(shí)隨地免費(fèi)播放，但我們?nèi)匀灰ㄥX去聽音樂會(huì)，看音樂家的現(xiàn)場表演。大多數(shù)藝術(shù)創(chuàng)作也是如此。雖然人工智能生成的內(nèi)容在某些情況下可能令人印象深刻，而且在某些領(lǐng)域甚至?xí)憩F(xiàn)得更突出，但它不能取代人類的創(chuàng)造力和表達(dá)。人工智能模型仍然受到訓(xùn)練它的數(shù)據(jù)和所使用的算法所限制，而且它們?nèi)狈θ祟悇?chuàng)造者的情感和直覺能力。Anthony Baker利用Midjourney創(chuàng)作的數(shù)字藝術(shù)此外，創(chuàng)作過程往往涉及一系列人類的經(jīng)驗(yàn)和情感，很難用人工智能來復(fù)制。因此，人工智能生成的內(nèi)容應(yīng)被視為人類創(chuàng)造力的補(bǔ)充，而不是替代。
Anthony Baker：我相信3D物體的創(chuàng)建和完整的視頻和CGI生成是AIGC的下一個(gè)前沿領(lǐng)域，以及能夠基于較小的人類思維和對(duì)話數(shù)據(jù)集來制定大型語言模型。雖然目前的很多AIGC令人印象深刻，并能創(chuàng)造出令人驚艷的內(nèi)容，但它們?nèi)匀蝗狈Υ罅康膫€(gè)性化和細(xì)微差別，而人類可以將其嵌入到創(chuàng)意內(nèi)容中，以適應(yīng)特定的環(huán)境、品牌、受眾、產(chǎn)品等等。Anthony Baker利用Midjourney創(chuàng)作的數(shù)字藝術(shù)
大型語言模型和一般的AIGC工具在使用數(shù)據(jù)方面的效率非常低。它們需要大量的數(shù)據(jù)和計(jì)算機(jī)能力來進(jìn)行訓(xùn)練，并且在小數(shù)據(jù)集的基礎(chǔ)上適應(yīng)某種行為模式的效率相當(dāng)?shù)停@使得它們?cè)谶@種情況下的效率非常低。這可能會(huì)在未來幾年迅速改變，并將由大公司和小公司的開源和閉源創(chuàng)新來推動(dòng)。我們將看到的另一個(gè)潛在趨勢是所有這些工具之間的協(xié)調(diào)和全方位渠道的創(chuàng)建。未來會(huì)出現(xiàn)許多公司和平臺(tái)，負(fù)責(zé)連接不同的AIGC，以提供完全集成的內(nèi)容制作解決方案，在那里你將能夠輸入一個(gè)提示，并得到一個(gè)完整的CGI視頻，以你想要的風(fēng)格，與你想要的虛擬角色，與你想要的標(biāo)題和語言。即使離完美還很遠(yuǎn)，仍然無法與完整的AAA級(jí)游戲或好萊塢電影競爭，但它將進(jìn)一步顛覆內(nèi)容和視頻制作。TOP君：對(duì)于AIGC未來的發(fā)展還有什么意見和分析？特別是在代理商方面，AIGC將如何影響代理商？
Anthony Baker：代理商需要謹(jǐn)慎面對(duì)AIGC發(fā)展。AIGC不是一種潮流或營銷炒作。人工智能已經(jīng)發(fā)展了70多年了。理論從根本上說是相同的，但我們現(xiàn)在擁有有效創(chuàng)造通用人工智能（general AI）所需的所有資源。社交媒體平臺(tái)是第一波人工智能驅(qū)動(dòng)的平臺(tái)，但在這上面做得并不出色。這第二代人工智能平臺(tái)將在我們的生活中產(chǎn)生更深更多的破壞性影響。Anthony Baker利用Midjourney創(chuàng)作的數(shù)字藝術(shù)我相信許多代理商將把AIGC工具看作是提高效率、獲得更好的利潤率和推動(dòng)克服過去3年由于疫情、烏克蘭戰(zhàn)爭、利率上漲和通貨膨脹造成的全球經(jīng)濟(jì)不穩(wěn)定的一種方式。許多公司正在研究利用AIGC使企業(yè)更賺錢的可能性，這是可以理解的。然而，我相信，那些投資于幫助員工、合作伙伴和未來新聘員工來接受、理解并更好地探索和利用AIGC工具的公司，從長遠(yuǎn)來看將是更成功的。ChatGPT、Stable Diffusion、Gen-2的領(lǐng)域?qū)?huì)發(fā)生非常迅速的變化。代理商與其考慮用AIGC工具來取代人力，不如考慮如何培養(yǎng)人才，讓他們更深入地了解這些工具，并學(xué)會(huì)使用AIGC工具來度過這波職場顛覆浪潮。創(chuàng)意和情感是人類經(jīng)驗(yàn)的內(nèi)在因素，但這并不意味著創(chuàng)意產(chǎn)出和過程不能改變和發(fā)展。我們只能根據(jù)我們目前的經(jīng)驗(yàn)和知識(shí)來判斷，所以我們需要賦予人們和我們自己更多了解和體驗(yàn)AIGC工具和平臺(tái)的能力，這樣才能超越我們目前的極限，發(fā)揮更大創(chuàng)造力。TopMarketing: What is the current level of creative creation in AIGC?
Anthony Baker: AI generated content has been in development for many years. However, the rapid progress in the development of large language models, massive neural networks and AI algorithms thanks to the availability of incredibly large data sets and computing power has enabled companies like Open AI, Midjourney, Google, Runway and Stability AI (to name just a few) to create impressive AIGC tools. Open AI’s GPT4, Runway Gen-2 and Midjourney v5 are recent examples on how far these tools have gotten. ChatGPT can now help developers to co-create and fix code, Midjourney is enabling a whole new generation of digital designers and artists to imagine and create digital creations in many cases far outreaching imagination, and Runway Gen-2 is poised to disrupt the video production process.It's no coincidence that Microsoft, Adobe and Google are rushing to integrate AIGC tools into their product suites, while a large group of tech world leaders and academics including Elon Musk (Tesla, Twitter, SpaceX), Steve Wozniak (Apple), Emad Mostaque (Skype) and Craig Peters (Getty Images) among many others published an open letter to ask AI development to pause for six months - arguably to give leaders and companies time to get their stuff in order and catch up.The state-of-play of AIGC is undeniable. Digital designs can be created out of text prompts in seconds, working code can be written with a few loose instructions, articles can be written in different languages and styles with a few chat messages and iterations, and video is about to be no exception. To think that these tools are not here to become ubiquitous in our daily lives and work is folly. Just look at folks without any industry involvement and tech knowledge interacting with ChatGPT for everyday chores and you will easily get how this is already becoming part of our society’s fabric.If you want any more assurance, just take a look at Hugging Face open source AI platform to see the hundreds of AI models, transformers, datasets and demos available for free to anyone out there. While most of these won’t be at the level of Open AI and Midjourney, they are not far, and they are completely open and accessible to anyone that cares to learn how to use them. Through history we have seen that anytime that both open source and closed source innovation reaches that much scale, big shifts are bound to happen.TopMarketing: How do creatives view the opportunities and threats posed by AIGC?Anthony Baker: It’s no surprise that such big shifts in technology and tools create a lot of different reactions. We have seen it happen over and over again throughout human history. Whenever technology innovation makes a big leap, it disrupts a lot of existing crafts, jobs, industries and processes. This is already happening, and jobs are going to be changed, destroyed and created by the AIGC developments, in a very rapid way over the next 2-3 years.People that do creative jobs, whether that’s creative ideas, creative design, creative copywriting, creative strategy or creative coding, are passionate about their craft. People have put so much effort into mastering their disciplines that it’s normal that if a tool threatens to do the same faster, cheaper and at a larger scale, it creates fear.What I have observed through online communities and social networks it’s quite interesting. There is a whole new wave of people that maybe didn’t have the creative job or craft to enable them to explore their full potential as artistic creators, and with AIGC tools they are now able to create things they never thought possible. There’s also the camp of creative creators that have always been at the forefront of innovation, and they embrace the exploration of AIGC tools as just another step in their pursuit of creativity and expression. And there’s the camp where they see these tools as a thread to their craft and income source. I would suspect that many in this field are worried about loss of income because of AIGC tools and the lack of time and resources (including lack of their companies commitment to training and upskilling).However, as any other technology tool, the opportunity lies in embracing these tools as co-creation partners. Coders and Designers should not be worried about these tools being better at them at doing mundane and repetitive tasks, they should be excited about how these tools could take their craft and creativity to levels never seen before.Think about this. Cars are faster than humans, but we still go and see people running. Music can be generated by computers (and has been for many decades) and we can stream it almost for free anytime and anywhere, but we still pay to go to a concert and see musicians play live. The same can be said for most artistic creations.While AI-generated content can be impressive and useful in certain contexts, it is not a replacement for human creativity and expression. AI models are still limited by the data they are trained on and the algorithms they use, and they lack the emotional and intuitive capacities of human creators.Additionally, the creative process often involves a range of human experiences and emotions that are difficult to replicate with AI. Therefore, AI-generated content should be seen as a complement to human creativity rather than a replacement for it.TopMarketing: What are the future directions and trends of AIGC?Anthony Baker: I believe 3D object creation and full video and CGI generation are some of the next frontiers for AIGC, as well as being able to personalise large language models based on smaller datasets of human thinking and conversations.While a lot of the current AIGC are impressive and can create impressive content, they still lack a lot of the personalisation and nuances that humans can embed into creative content to fit a particular context, brand, audience, product, etc.Large language models and general AIGC tools are very inefficient in how they use data. They need massive amounts of data and computer power to be trained, and are pretty inefficient to adapt to a certain behaviour of pattern based on small datasets, which makes them really inefficient in that context. That's probably going to rapidly change in the next few years, and will be fueled both by open source and closed source innovation by big and small companies alike.Another potential trend we will see is the coordination and creation of full stack pipelines between all these tools. Companies and platforms will emerge connecting different AIGC to offer fully integrated content production solutions, where you will be able to enter a prompt and get a full CGI video out of it, in the style you want, with the meta human character you want, with the caption and language you want. Even when far from perfect and still not competing with full AAA games or Hollywood movies, it will disrupt the content and video production even further.TopMarketing: Any other comments and analysis of future AIGC development? Especially on the agency front, how will AIGC impact agencies?Anthony Baker: Make no mistake. This is not a fad or a marketing hype. AI has been in development for more than 70 years. The theory is fundamentally the same, but we have now all the resources needed to effectively create general AI. Social Media platforms were the first wave of AI powered platforms, and we didn’t do great on it. This second generation of AI platforms will have an even deeper and more disrupting impact in our lives.I believe many agencies will see AIGC tools as a way to become more efficient, get better margins and push forward to overcome the last 3 years of global instability due to the pandemic, the war in Ukraine, Interests hikes and inflation. It’s understandable that many companies are looking into the possibilities to leverage AIGC to make businesses more profitable.However, I believe that it will be the companies that invest in helping their staff, partners and upcoming talent to embrace, understand and get better at exploring and leveraging AIGC tools will be the ones with more success over the long run. This field with ChatGPT, Stable Diffusion, Gen-2 is going to be changing very fast. Instead of thinking about what the agency is replacing with AIGC tools, the better bet is on how to nurture talent that will get deeper into these tools and learn to ride the disruption wave one tool at a time.Creativity and emotion are intrinsic to human experience, but that doesn’t mean that creative output and process cannot change and evolve. We are only able to judge based on our current experiences and knowledge, so we need to empower people and ourselves to know and experience more with these tools and platforms so we can be creative beyond our current limits.
本站僅提供存儲(chǔ)服務(wù)，所有內(nèi)容均由用戶發(fā)布，如發(fā)現(xiàn)有害或侵權(quán)內(nèi)容，請(qǐng)
點(diǎn)擊舉報(bào)。
