Asynchronous & Unreliable - Tech & AI Podcast
Sep 21, 2026 · 36 min · 10 segments
# Shownotes **Are AIs Learning Humanity's Better Nature? with Yanqing Cheng** Anne Currie talks with software engineering leader and Tollens.ai founder Yanqing Cheng about how large language models…
Anne CurrieHost
about, um, about long-form text writing being kind of intrinsic to the better angels of our nature, you know, better humanity.
There's Neil Postman, an American English professor, philosopher, very good, wrote very interestingly about the history of written text.
One of the things that they all point out is that when people write things long-form, really long-form, especially in books, and you iterate on it, you edit it, you iterate, you think about it, it's almost the best form of humanity is the book form of humanity.
I find it quite interesting that LLMs were trained on long-form text, and they are surprisingly nice.
The large language model without any post-training, so without RLHF and without reasoning training, they are capable of being nice and they are capable of mirroring any facet of humanity that they've ever read about.
They're just trying to describe the world that they're in because fundamentally the base unit of an LLM is text completion.
And then did I talk to you about this last time, the assistant persona and how it came about? no ah there's a whole very interesting rabbit hole here um how do you turn this large language model into how do you turn gpt into chat gpc no idea yeah oh there's a beautiful blog post about this um so you So the way that they initially did it is they said, you are a assistant.
And all of the nice adjectives that you could think of to describe the chat GPT you wanted.
about, um, about long-form text writing being kind of intrinsic to the better angels of our nature, you know, better humanity.
There's Neil Postman, an American English professor, philosopher, very good, wrote very interestingly about the history of written text.
One of the things that they all point out is that when people write things long-form, really long-form, especially in books, and you iterate on it, you edit it, you iterate, you think about it, it's almost the best form of humanity is the book form of humanity.
I find it quite interesting that LLMs were trained on long-form text, and they are surprisingly nice.
The large language model without any post-training, so without RLHF and without reasoning training, they are capable of being nice and they are capable of mirroring any facet of humanity that they've ever read about.
They're just trying to describe the world that they're in because fundamentally the base unit of an LLM is text completion.
And then did I talk to you about this last time, the assistant persona and how it came about? no ah there's a whole very interesting rabbit hole here um how do you turn this large language model into how do you turn gpt into chat gpc no idea yeah oh there's a beautiful blog post about this um so you So the way that they initially did it is they said, you are a assistant.
And all of the nice adjectives that you could think of to describe the chat GPT you wanted.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
