3 Years of ChatGPT
8 days ago
- #AI
- #Predictions
- #ChatGPT
- The author reflects on their initial skepticism and subsequent realization of ChatGPT's impact over three years.
- They discuss the evolution of AI from ML-assisted engineering to widespread daily use in personal and professional contexts.
- Key mantras include viewing AI as an amplification of intelligence and the importance of data quality and uniqueness.
- The author admits being wrong about AI's capabilities in audio interactions and coding automation.
- They were correct about the overrating of fine-tuning and benchmarks, emphasizing the value of vibe checks and expert labeling.
- One-year predictions include the rise of local AI inference stations and agentic e-commerce.
- Five-year predictions foresee coding becoming a specialized skill and the emergence of AI-native professionals in fields like medicine.
- Anti-predictions highlight that traditional industries and software architects will not see drastic changes soon.
- The author remains skeptical about breakthroughs in AI intuition and superintelligence in the near future.