The Reality Check Education AI Needs

Davos, Switzerland – Amidst the global excitement about artificial intelligence transforming education, Squirrel AI co-founder Liang Jing offered a sobering perspective at this year's World Economic Forum. Her message? The sector risks prioritizing flashy technology over genuine learning outcomes.

Beyond the Buzzwords

"Walk through any edtech conference today," Jing observed during her panel discussion, "and you'll see countless products claiming to revolutionize education with AI. But peel back the marketing layers, and most are just repackaged large language models."

The Chinese entrepreneur didn't mince words describing these offerings as "glorified textbooks" – systems that might deliver polished answers but fail to engage students meaningfully or adapt to individual learning needs.

Why Generic Models Fall Short

Education demands precision that general-purpose AI can't provide, Jing argued. She compared today's popular chatbot-style tutors to teaching assistants who merely recite textbook passages without understanding why students struggle with certain concepts.

"When we train human teachers," she explained, "we don't just teach them subject matter. We help them recognize confusion patterns, anticipate misconceptions, and adjust explanations accordingly. Our digital tools need that same depth."

The Path Forward: Vertical Specialization

The solution lies in what Jing calls "vertical large models" – AI systems specifically trained on:

• Millions of student interactions
• Detailed curriculum mappings
• Common learning obstacles
• Effective teaching strategies

Squirrel AI's approach exemplifies this philosophy. Their adaptive learning system reportedly analyzes over 100 data points per student interaction to identify precise knowledge gaps.

A Wake-Up Call for Investors?

The comments arrive as venture capital floods into education technology startups leveraging generative AI. Jing's critique suggests much of this investment might be chasing superficial applications rather than meaningful pedagogical breakthroughs.

The Davos audience included numerous tech executives and investors now facing uncomfortable questions: Are their products truly advancing learning? Or simply automating outdated teaching methods?

Key Points:

• Current edtech often repurposes generic AI without educational specialization
• True learning requires systems that understand pedagogy and student cognition
• Vertical models trained on educational data outperform general-purpose chatbots
• The industry must prioritize effectiveness over technological novelty