Skip to main content

What’s next in Generative AI?

Premium

We asked Gen AI professionals working in the industry: "What emerging skills or practices do you think candidates will need to master in the next 12–18 months?"

As of November 2025, here's what they told us.

In-depth understanding of evaluation metrics

It's no longer enough to know that evaluation matters. You need to understand the metrics deeply.

What this means:

  • Know when to use BLEU vs. ROUGE vs. human evaluation
  • Understand how to evaluate not just the LLM, but the entire system (retrieval quality, user experience, business outcomes)
  • Be able to design custom metrics for your specific use case
  • Track the right metrics at each stage: development, pre-release, and production

Why it matters: As AI products mature, the difference between good and great comes down to rigorous evaluation. Companies are moving past "does it work?" to "how well does it work, and how do we improve it?"

Agentic workflows

AI agents that can take actions, use tools, and complete multi-step tasks are becoming standard.

What this means:

  • Understanding how to chain multiple AI calls together
  • Knowing when to give AI access to tools (APIs, databases, code execution)
  • Designing workflows where AI makes decisions and takes actions autonomously
  • Managing reliability and error handling in multi-step processes

Why it matters: The next wave of AI applications won't just generate text, they'll complete entire workflows. Think AI that can research, draft, revise, and send an email without human intervention at each step.

Mastery of AI tools for productivity

AI-assisted development tools are becoming essential, not optional.

What this means:

  • Fluency with AI coding assistants (Cursor, GitHub Copilot, Claude for coding)
  • Using AI for rapid prototyping and iteration
  • Knowing when to rely on AI vs. when to write code yourself
  • Integrating AI tools into your daily workflow

Why it matters: Professionals who can ship 2-3x faster by effectively using AI tools will have a significant competitive advantage. This isn't about replacing human skills, it's about augmenting them.

Understanding dataset quality

"Garbage in, garbage out" is truer than ever with AI.

What this means:

  • Knowing what makes a high-quality training or fine-tuning dataset
  • Understanding how to create quality datasets at scale (not just perfect handcrafted examples)
  • Recognizing when data issues are causing model problems
  • Balancing dataset size with quality

Why it matters: Model performance is increasingly limited by data quality, not model architecture. Companies are realizing that investing in better data yields better returns than just using bigger models.

What do you think?

These are predictions from professionals working in Gen AI today. As the field evolves rapidly, we want to hear from you:

Next Lesson