Is Using AI to Have Celebrities Teach You English a Good Business?
-
Imagine Hollywood stars explaining English words in fluent Chinese—what a scene that would be?
Bilibili UP owner 'johnhuu' made it happen with AI. In this 1.5-minute video, the actress who played Scarlet Witch in The Avengers used clips from the show to explain in 'Chinese' how the same English word can cause misunderstandings in different contexts. So far, the video has garnered 3.96 million views on Bilibili.
Before this, 'johnhuu' also released a video titled 'This Is What Translation Without Dubbing Accents Looks Like,' using AI to show Taylor Swift speaking Chinese, which has now reached 3.52 million views. Compared to the original English videos, the AI has largely succeeded in 'restoring the original person' in terms of voice, tone, and accent, even synchronizing lip movements.
Currently, 'johnhuu' has created multiple AI-dubbed videos featuring celebrities, incorporating English teaching content to ultimately attract viewers to his English courses. This entertaining and educational approach to teaching English has garnered widespread praise from netizens.
What processes are involved in creating such AI-dubbed videos? What are the commercial prospects after combining AI with English teaching? Beyond 'johnhuu,' what similar accounts have emerged? And what potential risks and controversies do such accounts face? Learn English with AI in Just 3 Steps Using Your Favorite Celebrities
"Knowledge enters my brain in the strangest ways."
In the comment section of "johnhuu" videos, netizens often express such amazement. They are also surprised that these seemingly natural and fluent videos are actually created by AI, not narrated by the celebrities themselves. According to 'johnhuu's reply in the comments section, he stated that creating such a video usually requires meeting three conditions: text organization, voice cloning, and lip-syncing.
First is the text phase. Whether translating from Chinese to English or vice versa, it's essential to integrate local context and perform authentic semantic analysis to help the audience quickly engage with the dialogue.
In 'johnhuu's English teaching videos, he often starts by 'borrowing' a celebrity's voice to deliver a punchline, such as the word 'spoon' which can mean both a utensil and a type of embrace, then combines it with other film clips to unpack the joke: In the context of ordering food, 'spoon' meant 'I want a spoon,' but it was misunderstood by the son-in-law as hugging the father-in-law from behind.
The barrage area was filled with 'hahaha'.
Secondly, there's voice cloning. Recently, the article "10 Seconds to Have OpenAI's Sam Altman Dub Your Video? An 80-Year-Old Singer Goes Viral with This Tech" discussed related technologies. We attempted to recreate the voices of SpongeBob, OpenAI CEO Sam Altman, Messi, and Lazy Sheep using trending tools like ElevenLabs and CapCut's AI voice cloning.
To make the dubbing more vivid and emotionally varied, AI employs technologies such as "emotional intent recognition," "emotional feature extraction," and "natural language processing" to eliminate the cold, unnatural robotic tone.
After completing these two steps, the final stage is lip-syncing replacement. According to Top AI Player, websites such as HeyGen and LipDub can recognize speakers' lip movements to synchronize mouth shapes. Related reading: The Latest AI Voice Technology Goes Viral! Voice Cloning + Lip Sync—Now It's Really 'The Whole World Speaking Chinese'.
Combining the above steps, we find that the difficulty of creating star voice cloning + lip-sync videos posted by 'johnhuu' is not particularly high, but the content is cleverly designed and incorporates English teaching materials, leading to high view counts.
Currently, 'johnhuu' has amassed 1.71 million followers across platforms, with nearly 80% concentrated on Bilibili. With the productivity boost from AI, his video release frequency has recently increased. From January to February this year, he released 13 videos, more than triple the number from the same period last year. He mentioned in the Xiaohongshu comments that although AI-generated celebrity English teaching videos tend to have higher viewership, recent frequent use has led to platform restrictions. The video posted on March 2nd, which featured him personally explaining, showed a noticeable gap in view counts compared to other AI-generated celebrity videos.
Most of the above videos utilized AI synthesis technology, with seven surpassing one million views.
99-Yuan English Course: Business Opportunities and Risks Coexist Where there's traffic, there's potential business opportunity.
At the end of each video, 'johnhuu' includes the prompt 'Follow johnhuu, learn English happily,' and his Bilibili homepage introduction also features the URL of his Taobao store.
Currently, the 'johnhuu Teaches English' Taobao store has accumulated 17,000 followers. The store offers three products priced at 99 yuan each, with over 1,600 cumulative buyers and total sales exceeding 158,000 yuan. There are quite a few AI video creators like 'johnhuu' who blend English teaching with entertaining memes. Celebrities, song clips, and film/TV works can all become their material libraries.
For example, the ever-popular drama 'Empresses in the Palace' has been creatively adapted by the Xiaohongshu account 'Learning While Laughing.' They extract scenes from the show, use AI to transform them into scenarios like 'Yongzheng's English Private School,' and incorporate relevant knowledge points. A single post has received up to 84,000 likes.
Notably, this account leverages AI to repeatedly produce variations of the same video scene by only replacing the lip-synced dialogue, significantly improving publishing efficiency. Ultimately, 'Learning While Laughing' also directs traffic to sell educational courses. Unlike traditional dubbed videos, the AI-generated adapted videos mentioned above are more natural and fluent, thus attracting more traffic and attention.
However, since most of these video materials are taken from celebrities or films and TV dramas, combined with their commercial promotion model, they constitute a certain degree of copyright infringement, which is similar to the Bilibili ghostly recreation videos featuring celebrities. As a result, many videos by 'johnhuu' have been taken down.
Currently, while AI has allowed some educational bloggers to reap benefits, most large educational institutions have not used this method for customer acquisition due to the potential infringement risks. A new media director from a well-known educational institution informed us that the company's educational services have not extensively adopted AI. One reason is copyright concerns—"while the memes are amusing, they can't be used"—and another is the company's internal investment strategy, which remains cautious about AI technology.
However, the technology itself is neutral; the key lies in how creators use it.
On the legal and regulatory front, the government has also advanced relevant legislation. On August 15 last year, the Interim Measures for the Management of Generative Artificial Intelligence Services officially came into effect. The Measures stipulate that providing and using generative AI services must respect the legitimate rights and interests of others, must not endanger their physical and mental health, and must not infringe upon their portrait rights, reputation rights, honor rights, privacy rights, or personal information rights. Learning is inherently a tedious and monotonous process, but presenting it in an engaging format can enhance both interest and efficiency. As a result, some AI products have introduced features like English conversation practice, writing assistance, and dictation exercises to help users overcome challenges such as difficulty speaking and poor grammar.
For example, Doubao, an AI product under ByteDance, includes an "English Learning Assistant" module in its main interface, guiding users through the learning process. Additionally, user-created AI bots can serve as speaking partners for practice.
Beyond voice chat, Spark Language Partner, developed by iFlytek, supports video-based English learning with users, simulating real-life conversational scenarios more closely. In the AI era, every industry has the potential to be reinvented, and English education, with its inherent interactive dialogue nature, is poised to find even greater opportunities for growth.