Recent debates on X have , among other things, brought forth the purported underachievement attitude of the American education and children's upbringing. I have voiced my serious reservations regarding that point, to put it mildly.
Many of my own ideas on this matter have been heavily influenced by @DavidEpstein's book "Range". It is one of my favorite books on education and skill development in general.
1/13
David Epstein argues against the prevailing cultural narrative that specialization is the only path to success. Instead, he champions the idea of "range" — the benefits of having a broad set of experiences, skills, and knowledge. Epstein uses a variety of examples from sports, science, music, and business to illustrate how generalists often outperform specialists in complex and unpredictable environments. He suggests that learning broadly before (or even instead of) specializing can lead to greater creativity, adaptability, and success.
2/13
Key Points: 1. The Cult of Specialization:
Epstein critiques the "10,000 Hour Rule" popularized by Malcolm Gladwell, suggesting that early specialization isn't always beneficial. He points out that in many fields, late specializers often end up more successful.
3/13
2. The Value of Sampling:
Many successful people, including artists, scientists, and athletes, benefit from what Epstein calls "sampling periods" where they try different activities before settling on one. This broad exposure helps in developing a wide range of skills and understanding.
4/13
3. Learning to Learn:
Epstein emphasizes the importance of being a good "match learner" - someone who can learn from multiple fields and apply that knowledge broadly. This contrasts with "kind learning environments" where patterns repeat (like chess), where specialization can be more effective.
5/13
4. Interdisciplinary Insights:
The book highlights how breakthroughs often occur at the intersection of disciplines. Innovators like Nobel laureates often have a broad education or have worked in multiple fields before their big discoveries.
6/13
5. Delayed Specialization:
In fields like science, technology, and business, people who delay specialization and take a more wandering path through education or careers can end up more innovative and adaptable. This is particularly true in environments where problems are complex and solutions are not straightforward.
7/13
6. The Role of Grit vs. Flexibility:
While grit (perseverance) is celebrated, Epstein argues for "gritty quitting" where one knows when to abandon a path that isn't working. This flexibility can lead to better outcomes than stubbornly sticking to one narrow path.
8/13
7. Creativity and Problem Solving:
Generalists might not excel in routine tasks but are better at solving novel problems because they can draw from a wider array of experiences and perspectives.
9/13
8. Educational Implications:
Epstein suggests educational systems should encourage exploration rather than early specialization. He advocates for broader curricula that help students find their true calling later rather than forcing early specialization.
10/13
9. Workplace and Career Advice:
In careers, he suggests that having range can make one more employable in a fast-changing job market, where skills from one area can be valuable in another.
11/13
10. Sports and the Roger Federer Example:
Even in sports, where specialization is thought to be key, Epstein discusses how athletes like Roger Federer benefited from playing multiple sports in their youth, which helped develop a range of skills and physical capabilities.
12/13
Epstein's overarching message is that in an interconnected, unpredictable world, versatility and the ability to learn across domains are more valuable than ever.
13/13
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Today @SemiAnalysis published their extremely comprehensive, detailed, and honest report on performance comparison between @NVIDIA's H100/H200 GPUs and @AMD's MI300X.
1/9
On paper, MI300X has many advantages compared to the H100/H200, but in practice AMD's hardware is effectively nerfed by their catastrophically weak software. TL;DR: out of the box you will not be able to use MI300X for ML/AI training.
2/9
These findings confirm all the information that've heard over the years and has been privately shared with me. There is far, far more to running great ML/AI workloads, especially *at scale*, then having the experimental hardware specs that look great at your keynote address.
3/9
I've been holding off saying more about Google's purported quantum computing breakthrough until I read a bit more about it. (You should try doing something like that too!) It turns out, as I had suspected, it was waaaaay overhyped.
Yes, it's good science, but in terms of any kind of practical applications we are probably at least a decade away. Even then it will most likely be very specialized areas of application, like molecular dynamics.
2/4
A good rule of thumb is that quantum computers are really good at doing stuff that comes naturally to quantum mechanics. Which is *literally* all about randomness.
3/4
At the end of the last week @DarioAmodei, co-founder and the CEO of @AnthropicAI, one of the top AI labs, published a long essay on his vision of what an advanced AI in the upcoming years could potentially accomplish. There have been several other similar essays over the past few months from other top AI voices, but in my opinion this one is the most thoughtful and most detailed so far. It stirs away from many ideological squabbles that have become all too common these days, and provides ample citations that bolster his points, and help with further reading and self-guided research. Dario's own scientific background is in Biophysics and Neuroscience, and his takes on potential in those fields are particularly insightful.
1/7
Some key takeaways:
Biology and health
AI could drastically accelerate biological and medical progress, compressing 50-100 years of advancements into just 5-10 years. This could lead to the elimination of most diseases, significant extensions of the human lifespan, and greater control over biological functions like reproduction and aging. Key breakthroughs such as CRISPR, mRNA vaccines, and genome sequencing are examples of innovations AI could multiply, transforming human health.
2/7
Neuroscience and mind
Just like with the biological research, AI stands to fast-track neuroscience research, possibly curing most mental illnesses and boosting cognitive and emotional abilities in just 5-10 years. AI is set to improve mental health treatments, discover new drugs, and help with everyday cognitive challenges, making life more fulfilling for everyone.
3/7
.@NVIDIA has just announced TensorRT-LLM, open source software designed to accelerate Large Language Model inference on H100s.
1/7
This software has been developed though close collaboration with many leading AI companies, such as @Meta, @anyscalecompute , @cohere , Deci AI, @Grammarly, @databricks and many others.
2/7
Depending on an LLM and the use case, TensorRT-LLM can provide up to 2X speedup in performance.
3/7
By now most of us well aware of transformer-based large language model capabilities and, in many instances, failures. The failures in particular can seem extremely head-scratching, as they often involve the kind of mental reasoning that even a young schoolboy cold do.
1/6
A new paper tries to investigate the nature of these failures, and understand the limits of LLM-based reasoning. It seems that the failures primarily arise from the tasks with low in-domain knowledge and high compositional complexity.
2/6
This analysis can also open the doors for new approaches to AI for much more comprehensive problem solving skillsets.
3/6
Very exciting news - Python is now available in the official version of Excel! Excel is the most widely used analytics tool in the World, and Python has become the most popular programming language for Data Science and Machine Learning tasks.
1/4
It is a very intuitive and easy to learn programming language. The merger of these two tools will open new opportunities and use cases.
This merger is the culmination of years-long effort and collaboration between Microsoft and the open source Python community.
2/4
Python in Excel is right now available as a Public Preview, and Python calculations run in the Microsoft Cloud. Many of the popular Python Data Science and Machine Learnign libraries are currently supported, such as Pandas, Matplotlib and scikit-learn.
3/4