Lyric's Blog

o1 pro remains top coding model. Break complex tasks into simple ones with clear boundaries. Humans excel at intermediate engineering tasks. High-quality documentation crucial for AI comprehension.

AI is the magic source of the age.

The AI project has stopped due to intense competition and limited commercial scenarios. Prompt engineering, AI replacements for classical functions, RAG, and tool products are some viable applications. User experience and competitive advantage in AI technology are crucial. The development cost and quality of AI applications are challenging. The decision to target the consumer or enterprise market depends on market size, revenue model, customer behavior, market response, marketing strategy, and product features.

The author explains how AI has significantly increased their efficiency in coding, writing copy, and drawing illustrations. They describe how AI can generate code for mundane tasks, help with writing technical documents, and provide illustrations without copyright issues. This has resulted in significant time savings and increased productivity.

Meta's AI competitor Llama 2 can now be run on AMD Radeon cards with ease on Ubuntu 22.04 Jammy Jellyfish. The process involves downloading the Llama 2 models, compiling llama.cpp, testing it, and enabling GPU acceleration. The GPU acceleration reduces waiting time and improves performance.

Programming with AI involves interacting with OpenAI's language models and requires understanding the difference between pure and impure functions. Impure functions pose challenges in code reuse and testing. The use of agreed-upon standards, such as specifying the expected output format, can help standardize AI outputs. Using English prompts is recommended due to the dominance of English in training data and cost considerations. TLDR AI programming involves completing user input and processing it through language models.

Blockchain and AI are facing increased regulatory pressure from governments. Cryptocurrencies are being targeted by regulations and restrictions, but they continue to gain support and demonstrate their strength. AI technology, dominated by big companies, is criticized for its power and potential impact on individual freedom. Open source is suggested as the best approach to ensure AI's freedom and prevent centralized control. Freedom comes at a cost, but as technology develops, the price decreases. Combining Large Language Models (LLMs) with blockchain can provide freedom of access and personal security. The existence of 'super individuals empowered by technology' must be accepted. Freedom and flexibility can be regained by moving away from centralized cloud services. Reduction in costs and restoration of engineering flexibility can be achieved.

This generation's AI technology has reached a level where it can replace many human tools in terms of productivity. Programmers and designers cannot be directly replaced by AI, but AI can assist them in their work. However, programmers may be indirectly replaced due to increased efficiency. High-level designers cannot be directly replaced, but tool-level designers like illustrators may be replaced. Senior product managers cannot be directly replaced by AI.