AI is increasingly used in language learning, as a growing number of students around the world are turning to large language models (LLMs) or AI-powered Chinese-learning apps and platforms, which promise to teach the language in record time. At first it looks great: shiny graphics, quick wins, everything going smoothly.
But there is always a catch, which is why The Chairman’s Bao (TCB), the online platform which publishes graded Chinese readings built on real news for people learning Mandarin, has made a choice that might sound almost radical: TCB’s Chinese lessons are not generated by AI.
They are created by humans, checked against HSK standards, and recorded by native speakers.
A quick clarification before we go further. In this article, “AI” refers to generative AI and large language models (LLMs) used to create lesson content: texts, grammar explanations, exercises and audio.
This does not mean that TCB is “against AI” in absolute terms. Artificial intelligence can have its place in internal operations, for example, to support the team’s daily tasks. However, TCB does not use AI to generate the educational content for its Chinese lessons.
And this article will explain why.
What Do Human-made Lessons Give You that AI Can’t?
What are you getting when a TCB lesson says “human-made”? Three things:
Accountability. Every article passes through editorial review by people who understand both the Chinese language and the HSK levelling system. If something is wrong, a human is responsible, and a human fixes it.
Authenticity. TCB lessons are built on real-world news and topics and written to sound like something worth reading and discussing at dinner, not something created to fill a template.
Real voices. Native speakers recorded at speeds appropriate for the level with the rhythm, warmth and natural phrasing you will actually hear in China.
| AI-generated lessons | The Chairman’s Bao | |
| Who creates the content | A language model predicting likely text | Human creators and editors |
| Accuracy | Plausible errors: invented grammar rules, context-free vocabulary | Every article is human-checked for accuracy against HSK standards |
| How it sounds | Statistically average Chinese: flat and generic | Authentic, idiomatic Chinese worth talking about |
| Audio | Synthetic voices with no warmth, rhythm or regional colour | Native speakers recorded at level-appropriate speeds |
| Accountability | Nobody is responsible when it is wrong | A named editorial team stands behind every lesson |
| What it trains | Talking to a machine | Talking to people |
In short: TCB’s competitive advantage is human-created, news-based graded lessons, checked against HSK standards, with native-speaker audio and a named editorial team accountable for accuracy.
Why Does TCB Choose Human-Created Content Over AI?
Following the appearance of AI lesson-generation tools across the industry in 2022-23, the TCB editorial team ran them through the same review process every TCB article must pass: grading checks, grammar verification, native-speaker review.
The test lessons failed, but why?
-
AI is not accurate enough, with frequent mistakes during lesson creation.
AI-generated Chinese reads smoothly. Yet the tests of TCB kept finding grammar explanations that were talking about rules that didn’t exist, vocabulary that wasn’t put into a context that made sense, and example sentences that were correct in a technical sense but awkward in real life… the written equivalent of someone smiling at the wrong moment in a conversation.
Beginners would not see these mistakes and more experienced students would question what they know, then learn and do it again and again, spending months building on sand.
-
AI produces a robotic tone and lacks applied authentic content.
In TCB’s tests, AI tended to produce statistically average, flattened Chinese… Language is human and it is meant to be fun, a little messy and shared!
If you want to turn vocabulary into conversation, you have to study things you can actually talk about: real news, real culture, real stories with idioms, humour, tone and cultural weight.
When you talk to a computer by yourself, it becomes a one-way transcript of vocabulary with little to no authentic language, which makes you sound like a phrasebook rather than a real person.
TCB wants students to have the whole experience.
-
AI audio is dreadful, with no accents or natural tone of voice.
Another interesting result from the tests was the AI-generated Mandarin audio being dreadful: it lacked personality, natural rhythm and regional colour.
You learn to speak the way your models speak, so if you train your ear on robotic audio, you will speak like a robot. And learning to speak like a robot will not serve you well when you are randomly dropped into a conversation in Chongqing.
This confirmed the TCB team’s decision to keep every lesson recorded by native speakers because your ears deserve better than a simulation.
-
The environmental and ethical impact of AI is abhorrent.
Creating educational content at scale with huge language models has a significant environmental cost. The TCB team does not believe the expense of its lessons is justified when the alternative is superior in every aspect, namely, human editors who create content for humans. There is also an ethical dimension to the employment of automated content and opaque models, which raises concerns about the worth of human labour, editorial accountability, and data use. TCB prefers an approach in which a specific team of people creates the information that students read and hear.
There Are No Real Shortcuts to Learning Chinese
AI often presents a false idea of a faster and easier learning process, framing language acquisition as something that can be quickly “hacked.” In reality, there are no shortcuts to learning a language. It is a slow, but deeply rewarding process.
In the academic world, there is actually a term for this: “metacognitive laziness” which means we get so used to letting technology do the heavy lifting that we stop doing the mental work ourselves. But if you lean on an AI model too much, you lose that productive struggle, the exact friction they need to really grow. It makes it far too easy to check out, instead of really getting your teeth into the language.
In addition to this, AI-generated texts and audios significantly decrease student engagement. One of the biggest things the TCB team prides itself on is keeping students genuinely engaged. Talking on your own to a computer is isolating, whereas better student engagement is one of the core benefits of TCB. By providing authentic, human-created content, TCB keeps learners highly engaged and builds confident language learners, not just people who know how to talk to an LLM.
TCB and AI: the Bottom Line
TCB definitely isn’t anti-technology. Tools that make processes smoother are welcome, and AI absolutely helps the team out behind the scenes. But when it comes to the actual Chinese you are studying? The Chairman’s Bao firmly believes that has to come from real people, not predictive models. Technology is a great assistant, but it never gets to be your teacher.
If you want Chinese you can actually use, real news, real voices, real conversations, that is exactly what TCB is here for.
Frequently Asked Questions
Does TCB use AI anywhere on the platform?
All TCB articles, grammar lessons and audio are made by people. The things you learn from are never made by AI. In the background, The Chairman’s Bao uses AI tools for internal work. One rule applies: AI is there to help the team, never to act as your teacher.
Is AI-generated Chinese really that inaccurate?
Often enough to matter. AI tools also produce “average” Chinese: technically fine, but flat and unlike how native speakers actually speak and write.
Can I still use AI tools alongside TCB in my studies?
Of course! TCB’s advice: AI should check and support your work, not do your work. The moment it starts thinking for you, your Chinese stops growing.
Why does human audio matter if the text is the same?
Because you learn to speak like your models. Human recordings capture the rhythm, tone, accent, and warmth of human speech, making a difference between being understood and sounding natural.









