Prioritize Deep AI Expertise Over Early Startups

Paul Graham recommends that 17-year-olds focus on building LLMs from scratch to create a strong knowledge base for future startups. Yann LeCun adds that they should investigate why AI struggles with physical tasks and seek architectures beyond current LLMs. Together, they advocate for deep technical exploration over early entrepreneurship.

Key Points

• Paul Graham advises 17-year-olds to master building and training LLMs from scratch rather than starting a business immediately.
• Deep foundational knowledge of AI is expected to yield much better startup ideas in the long run than early entrepreneurship.
• Yann LeCun highlights the gap between AI's cognitive abilities and its inability to perform simple physical tasks.
• LeCun encourages the study of new architectures beyond LLMs to solve the problem of physical task execution.
• Both experts emphasize the importance of academic and technical depth over superficial application.

Sentiment

Skeptical yet nuanced; while many find the vocational aspect of the advice impractical, they generally support the underlying philosophy of learning through low-level implementation.

In Agreement

• Building from scratch develops deep technical intuition and first-principles understanding of the 'substrate' of the future.
• Similar to building an OS or a browser in the past, the exercise makes you a better engineer even if you don't build that specific product professionally.
• Resource constraints on commodity hardware can force teenagers to find clever, innovative ways to optimize models.
• Small-scale models (sub-billion parameters) are actually feasible to train on modern consumer GPUs or through affordable cloud rentals.
• Understanding the 'black box' of AI provides a competitive edge over those who only know how to use high-level APIs.

Opposed

• The advice is out of touch with the reality that 'real' LLM engineering requires millions in capital and compute.
• The job market for model training is tiny and dominated by a few big labs that require PhDs and specific conference publications.
• Most companies do not need model training; they need 'software carpentry' like infrastructure, deployment, and fine-tuning.
• The 'Large' in LLM means that toy projects at home may not translate to the actual problems faced at scale in the industry.
• 17 is an age where broad exploration or fundamental CS/EE skills might be more valuable than hyper-focusing on a potentially transient tech trend.