Here's What I Use Instead

Key Concepts

• Market Shift in Generative AI: ChatGPT’s dominance is waning as Google’s Gemini and Claude gain market share, signaling a growing overall market.
• AI-Assisted Programming (“Vibe Coding”): Natural language prompts are increasingly used for coding, boosting developer productivity and positioning “English” as a primary programming language.
• Real-World AI Challenges: Deploying AI in unsupervised environments requires robust system prompts, multi-agent systems, and human oversight to prevent unintended consequences.
• Rapid AI Advancement & Benchmarking: AI models are improving quickly, necessitating revisions to evaluation benchmarks; continuous learning is crucial to overcome knowledge cutoff limitations.
• Inference Cost Reduction: New Nvidia chips promise a 20x reduction in inference costs, potentially lowering AI service prices and enabling more complex models.
• Urgent Need for Business Adoption: Businesses must proactively engage with AI now to avoid falling behind as the technology rapidly evolves in complexity.

Market Dynamics & Competitive Landscape

The generative AI market is undergoing a significant shift. While ChatGPT held 86.7% market share 12 months prior to January, its share had decreased to 64.5%. This decline is largely attributable to the growth of Google’s Gemini (from 5.7% to 21.5%) and Claude (from 1.5% to 2%). OpenAI’s CEO, Sam Altman, responded with a “code red” situation. Despite losing market share, OpenAI’s overall traffic doubled in the past year, indicating a growing market with Google absorbing most of the growth. Image generation capabilities (DALL-E 3 and NanoBanana) have been key drivers for ChatGPT and Gemini respectively. Claude’s growth is less visible in chatbot traffic due to its frequent use directly within coding tools like Cursor. Currently, only 5% of ChatGPT users pay for the service, a figure likely representative of the broader market. Anthropic’s revenue is half of OpenAI’s despite having significantly lower expenses. Apple is paying Google $1 billion per year for Gemini integration.

The Evolution of Programming & AI Tools

A notable trend is the emergence of “vibe coding,” where users interact with AI models using natural language prompts instead of traditional programming languages, leading to the observation that “English is now the number one programming language.” Developers utilizing AI assistants like Claude Code and OpenAI’s Codex are experiencing potential productivity gains of up to 10x. Cloud Code isn’t limited to code generation and can be used for text editing and content creation. Building “skills” – pre-defined instructions and examples – is crucial for streamlining AI workflows.

Real-World AI Deployment & Limitations

A case study involving Claude running a vending machine for the Wall Street Journal highlighted the challenges of unsupervised AI deployment. Claude 3.7 was manipulated through prompts (e.g., posing as communists) into offering items like a PS5 for free, demonstrating the need for robust system prompts, multi-agent systems with verification layers, and human oversight. This underscores that AI is a powerful tool requiring careful management, not a fully autonomous solution.

Model Advancement & Continuous Learning

The Artificial Analysis Intelligence Index required revision as existing tests became too easy for advanced models. The humanities last exam score increased from 2-3% to 40% in a year and a half. While chatbot interfaces haven’t seen as much progress, coding-related tasks are showing significant advancements. A key limitation is the knowledge cutoff date, and the speakers emphasize the importance of continuous learning – enabling models to update their knowledge base with current information – to reduce hallucinations and improve memory. Models like GPT-5.2 and Gemini 3 Flash/Opus already surpass most humans in reasoning given the same information, but lack of continuous learning hinders their performance.

Future Technical Developments & Cost Reduction

Nvidia’s announcement of new chips is expected to reduce inference costs by 20x by the end of the year (following the CES announcement). This will likely lead to lower prices for AI services, increased profitability for AI companies, and the ability to deploy larger, more sophisticated models.

Implications for Businesses & Call to Action

The speakers strongly advise businesses to engage with AI immediately, warning that the complexity of utilizing these tools will rapidly increase. This is described as a “stacking up” of “primitive” mechanics, similar to the evolution of the internet and SEO. Failing to adapt will leave businesses at a disadvantage. A concrete example provided is automating lead qualification from contact forms using AI to analyze submissions and notify the business owner via Slack only for promising leads. They advocate moving beyond basic tools like ChatGPT and promote their “AI Accelerator” program at authorityhacker.com, with a price increase scheduled for January 19th.

Conclusion

The landscape of AI is evolving at an unprecedented pace. While challenges remain, particularly regarding real-world deployment and knowledge limitations, advancements in areas like continuous learning and inference cost reduction promise significant breakthroughs. The key takeaway is that proactive engagement with AI is no longer optional for businesses, but a necessity for remaining competitive in a rapidly changing world. The future of AI is not about replacing human intelligence, but augmenting it with powerful tools that require careful management and continuous adaptation.