Welcome to the book summary of AI 2041. In this book, Kai-Fu Lee, the former president of Google China and bestselling author of AI Superpowers, joins forces with celebrated novelist Chen Qiufan to imagine our world in 2041 and how it will be shaped by AI. In ten gripping short stories, set twenty years in the future, Lee and Chen introduce readers to an array of eye-opening characters and situations. By gazing toward a not-so-distant horizon, AI 2041 offers urgent insights into our collective future—while reminding readers that, ultimately, humankind remains the author of its destiny.
This chapter starts with a story of the twin brothers “Twin Sparrows” and their personal AI companions. The author led us to a world where education has been truly equalized with the advance of NLP. It’s actually surprising to see just 2 years after the release of the book, we have witnessed so much development of AI especially in the field of NLP.
Contents
- Supervised and self-supervised NLP.
- Will AI have general intelligence when it masters our language?
- The future of education in the AI era.
NLP
Here is the timeline of some milestone achievements in the field of NLP.
2017 - Google invented transformer;
2020 - OpenAI released GPT-3 (Generative Pre-trained Transformers) which was trained with 45 terabytes of text and 175 billion parameters;
2021 - Google released a new language model with 1.75 trillion parameters (10X GPT-3).
2023 - OpenAI released ChatGPT (GPT-3.5) which marks the letting human interact with AI.
GPT - Since then, all companies are running full speed to participant in the AI game.
The author mentions that the most exciting aspect of GPT model series is that it will become a new platform, or a foundation on which domain-specific applications could be built quickly. At certain time, with the improvement of GPT, it will become platforms like Windows and Andriod where tens of thousands of developers will create amazing apps leveraging it.
[BackToReality] - At OpenAI’s first Dev Day in Nov 2023, they announced the plan of the GPT Store, just as the author predicted.
Example applications:
- Conversational AI (chatGPT): Why would this be a big deal? Because it’s capable of offering help 24/7, which humans often cannot.
- Next-generation search engine, it will shift how search engine work and answer any questions.
Another big discussion is - is this AGI (Aritificial General Intelligence)?
Turing test, AGI and consciousness
Central to human intelligence are the abilities to reason, plan, and create. One critique of deep-learning is that they will never have human intelligence. Skeptics believe AI should interact with human with different constrains and in the comning decades, these various theories will be put to the test and we will see.
The author doesn’t believe deep learning will become AGI (Aritificial General Intelligence) and match human intelligence by 2041. He states, to achieve human intelligence, we need a dozen more breakthroughts like deep learning, which was the one and only biggest breakthough in over 60 years.
The author suggests that we should focus on developing useful applications suitable for AI and seek to find human-AI symbiosis, rather than obsess about whether or when deep-learning AI will become AGI.
AI in education
The flaws of today’s education is that it’s one-size-fits-all yet we know each student is different. However, it’s expensive and cannot be scaled to poorer regions with a reasonable student-to-teacher ratio. AI can play a major part in fixing these flaws and transform education.
Perhaps the greatest opprtunity for AI in education is individualized learning. A personalized AI tutor could be assigned to each student which made learning more fun as all the learning can be customized based on the interest of the studnets.
In addition, human teachers’ routine work like grading homework can be replaced by AI. Human teachers can focus on building emotional intelligence creativity, charactor, values, and resilience in students.
With AI taking over significant aspects of education, basic costs will be lowered which will allow more people to access education. It can truly equalize education by liberating course content and top teachers from the confines of elite institutions and delivering AI teachers that have near-zero marginal cost.
Here conclude the end of this chapter, how would you vision the our world with AI in 20 years?