Greetings, curious mind! đź§
Lately, there's been a fascinating shift in the AI world. Many self-titled GenAI engineers aren't building traditional AI/ML models like the old days. Instead, GenAI has moved into the realm of API applications and SDKs, leaning more toward software engineering than deep AI research.
But, as we dive deeper—exploring workflows, plugins, and personas—we start brushing up against ideas like tuning and retrieval-augmented generation (RAG). Here's the kicker: these new methods don't demand deep statistical expertise.
A New Era of AI
I’ve had conversations with GenAI engineers, and it’s striking how often traditional AI terms like R-squared, p-values, or concordance/discordance don’t even come up. It begs the question: Do they need to?
AI and data science have undergone dramatic shifts. Back in 2017-18, neural networks started to edge out statistical methods. Terms like learning rates and batch sizes took center stage. Now, with GenAI, we’re seeing another evolution—not in the core technology itself, but in the abstraction layer.
Pre-built Tools vs. Custom Models
Most big tech companies have their own modified packages, reducing the need to build models from scratch. For instance, if you're tasked with solving a time-series problem, how many data scientists (even from Big Tech) would develop custom ARIMA, ARIMAX, or LSTM models? Few.
Instead, most opt for pre-built solutions like Prophet, DeepAR, or even AutoML Forecast. And that’s perfectly fine.
The Key Takeaway: It’s About Solving Problems
At the end of the day, what truly matters is solving business problems and delivering real results. Technology is a tool, not the end goal itself. The value doesn't come from debating who knows more about AI technicalities—it comes from driving outcomes.
The future of AI is clear: it’s not about every engineer building models from scratch but about knowing how to apply the right tools for the right job.
So, as the world of GenAI continues to evolve, it’s not less exciting—it’s just changing direction. Adapting, learning, and using these tools effectively are the new keys to success.
Stay tuned as we continue exploring the GenAI landscape. The journey has only just begun! 🚀
Cheers to innovation and solving real-world problems! 🥂