AI Still Cannot Fully Replace Humans: The Complementary Progress of Intelligent Technology and Human Wisdom
-
Since the emergence of ChatGPT, the industry has witnessed a surge in AIGC applications: on the user side, from Wenxin Yiyan to Miaoya Camera, new tools have emerged one after another, quickly becoming hits on social media; on the enterprise side, numerous 'AI+SaaS' services have appeared, claiming to replace a large amount of repetitive work in management and operations.
For a time, discussions about AI replacing copywriters, designers, live-stream hosts, and other professions became rampant, sparking widespread anxiety about which jobs would be the first to go.
However, after several months of marketing and live-streaming practices, people have gradually uncovered the shortcomings of AIGC. Even Jasper, an AI unicorn specializing in large models, recently announced layoffs.
Has the bright future of AIGC not yet fully arrived?
1. The Future—Wait, It's Not Here Yet
At least in the field of AI live-streaming, the future where AIGC dominates everything is still far away.
Ricky, Vice President of Baozun E-commerce, shared insights on AI live-streaming practices at the 'AI Empowers New Consumption' summit held by Shequ Neighbors on July 20.
Currently, AI live-streaming is divided into fully automated broadcasts and AI-assisted human broadcasts. The former is entirely handled by AI, from the host's appearance to the content, while the latter uses technology to extract and simulate a real person's appearance and movements, with the text content still delivered by AI—essentially a 'masked' broadcast.
Ricky pointed out that the advantages of AI live-streaming include: (1) reducing costs by nearly half compared to human hosts, (2) operating 24/7, (3) accurately responding to every consumer comment without missing any, and (4) avoiding risks like verbal mistakes during live broadcasts.
However, a comparison of GMV data between AI live-streaming, AI-assisted human live-streaming, and fully human live-streaming quickly dispels the idea of completely replacing humans with AI—human hosts, with their richer emotions and genuine interactivity, achieve an overwhelming advantage in driving consumer purchases.
Thus, Ricky believes that, for now, AI live-streaming can only serve as a supplementary tool to human hosts and cannot fully replace them.
Similarly, many domestic education products claiming to feature AI tutors, such as Duolingo, Liulishuo, Zuoyebang, and Yuanfudao, may surpass humans in response speed for problem-solving and teaching, offering personalized learning plans. Yet, they still fall short in emotional connection.
In fact, emotional value is indispensable to consumers.
Coincidentally, in text and image creation, many companies are realizing that AI isn't as easy to use as expected.
Since ChatGPT's debut, Li Rui (pseudonym) has stopped receiving copywriting optimization gigs from an advertising consultancy—a previously stable side job covering brand introductions, product advantage summaries, and usage guidelines.
Upon inquiry, Li Rui discovered that the company's employees were diligently "training" ChatGPT, hoping AI could eventually replace his job. Though frustrated, he recognized this as an inevitable trend—after all, AI is significantly cheaper than human labor and theoretically inexhaustible.
Unexpectedly, three months later, the company approached Li Rui again to resume their collaboration.
The reality proved challenging: For AI to generate coherent text or images, it first requires massive amounts of accurate input data. Otherwise, the output might be completely irrelevant. Even when AI completes a creation, substantial human effort is still needed for editing and refinement...
Additionally, ChatGPT's American origins make it less adept at handling Chinese linguistic nuances, while domestic alternatives like Wenxin Yiyan and Xunfei Xinghuo remain immature—often producing lengthy passages of confident-sounding nonsense that require human rewriting.
"Using AI proved too cumbersome—they concluded it's easier to keep working with me," Li Rui shrugged, unable to hide his satisfaction at outmatching AI.
This pattern extends across AIGC applications. An illustrator shared with New Retail Business Review that while initially fascinated by AI tool Midjourney, she gradually found its works possessed an "uncanny beauty" that eventually lost its appeal.
Her illustrator peers generally view AI art with hostility, seeing it as threatening professional survival and disrupting the creative industry's ecosystem.
If AI were a sword, achieving true harmony between human and machine remains distant.
2. A Double-Edged Sword
Many products now incorporate AI primarily as a marketing gimmick—more like seasoning added to attract consumers.
AI's hype is undeniable. Creating a digital avatar on the newly launched Miaoya Camera mini-program initially required over 8 hours of waiting time, which still exceeded 2 hours a week later—a popularity any brand would envy.
But Miaoya Camera's fame likely won't last. We've witnessed countless short-lived viral products that "explode overnight, get copied by morning, and lose favor by evening."
These products serve value by showcasing new technologies and helping industries identify their boundaries.
Technology is a double-edged sword, and AIGC can backlash against brands and platforms.
Foremost are data security and privacy concerns.
The most criticized aspect by consumers is AIGC services, particularly AI customer service. Nowadays, you might receive three or four greeting calls from AI customer service in a single day, where they use monotonous tones to ask if you need a line of credit, a credit limit increase, or if you're interested in a certain product, or if you haven't visited a flagship store in a while...
Sometimes, if you respond a bit slowly, they ruthlessly repeat their script, rendering any attempt to interrupt futile—the only solution is to hang up.
"I don't have much goodwill toward brands or platforms that use AI customer service—it feels insincere," a frustrated consumer complained to New Retail Business Review. "I feel like they're monitoring my every move, presumptuously offering so-called services I need, making me feel like my personal life is exposed..."
Moreover, when actually needing help, these AI services "only regurgitate the same phrases, not only failing to solve my problem quickly but also adding to my frustration."
Consumers also worry about fraudsters using AI to synthesize their portraits or voices to scam their relatives. "Our portraits, voices, fingerprints, and other information are stored on internet platforms. If this data is exposed, the consequences could be dire," the consumer expressed.
Next is the issue of copyright and ownership.
In May this year, someone created an AI version of singer Stefanie Sun (孙燕姿) to perform Jay Chou's (周杰伦) songs. Podcast host Pan Luan and ACE Virtual Singer founder Guo Jing discussed copyright ownership—the lyrics and music copyright should still belong to Jay Chou, while the AI Stefanie Sun is merely an "instrument" performing the song. The creator of this "instrument" cannot release songs under "Stefanie Sun's" name, as it would infringe on her rights.
Also in May, the Hangzhou Internet Court ruled on China's first virtual digital human infringement case, finding a Hangzhou-based company guilty of copyright infringement and unfair competition—it had used another company's digital human videos for profit without authorization. The company was ordered to pay 120,000 yuan in damages.
To mitigate such risks, Douyin (TikTok's Chinese counterpart) took a drastic step—banning fully automated AI livestreams and subsequently issuing AIGC platform guidelines, requiring AI-generated content to be clearly labeled to distinguish it from human-created content.
Local lifestyle platforms like Dianping have also begun cracking down on AI-generated reviews and promotional content.
In fact, the business model of AIGC itself remains unclear and is still in a money-burning, land-grabbing phase. For example, the company behind the Jasper model recently announced layoffs, while OpenAI, the creator of ChatGPT, is also struggling. According to media reports citing insiders, due to ChatGPT's development costs and hiring key employees from Google, OpenAI's losses roughly doubled in 2022, reaching approximately $540 million.
Many misunderstand the purpose of AIGC. AIGC was born to assist humans, not to replace them.
ChatGPT and Midjourney essentially generate text and images based on big data, replacing repetitive and mundane tasks, allowing humans to focus on more creative, decision-making, and AI-unimaginable work.
"AIGC is only making choices, not creating," said Pan Luan, host of Random Reading.
Han Qing, founder of AI Talk, believes that AIGC practitioners are currently divided into two camps. The first is the imitation camp, which focuses on using AI to produce content that closely resembles human output. However, he considers this only the initial stage of AIGC, which is what most in the industry are currently doing.
The other camp is the creativity-stimulating camp, which leverages AI technology to explore new possibilities in human art, creating content that surpasses human imagination. However, reaching this level may still require some time.
After the industry's initial excitement, AIGC has entered a relative 'cooling-off period,' with more stable development and increasingly clear industry standards and legal boundaries.
Hu Shiwei, co-founder of Fourth Paradigm, stated: "A technology often faces skepticism or fades away after its initial hype because it fails to meet the public's expectation that 'technology should bring significant changes to production and life.' However, it is often during the quiet period that the technology truly delivers value and solidifies its impact." He also noted that the trend of AI creating value will "rise wave after wave," with "the intervals between peaks gradually shortening."
AIGC will undoubtedly continue to evolve, but its influence will become more subtle, permeating everyone's lives and becoming an omnipresent force. People will gradually adapt to working and living in collaboration with AI.
Are you looking forward to it?