---
id: "claim-genai-lacks-depth"
type: "claim"
source_timestamps: ["¶46", "¶47", "¶48", "¶50"]
tags: ["artificial-intelligence", "generative-ai", "limitations"]
related: ["open-question-ai-data-privacy", "claim-ai-productivity-enabler", "entity-indra-nooyi"]
speakers: ["Indra Nooyi"]
confidence: "high"
testable: true
source_url: "https://hbr.org/2025/10/innovating-at-the-core-and-for-the-future"
source_title: "Innovating at the Core—and for the Future"
sources: ["futures"]
sourceVaultSlug: "hbr-seg-futures"
originDay: 2
articleStem: "hbr-cl-91-innovating-core-and-future"
sourceUrl: "https://hbr.org/2025/10/innovating-at-the-core-and-for-the-future"
sourceTitle: "Innovating at the Core—and for the Future"
---
# Generative AI currently lacks depth, perspective, and currentness

**Confidence: high. Testable: yes.**

Based on a personal test — Nooyi had her staff use GenAI to answer questions drawn strictly from her own past interviews — she concluded that while the output sounded *incredibly intelligent* on the surface, it ultimately lacked depth, practical application, and current perspective. She calls it a **'starter pack'** and insists human intervention remains critical for true insight. This complements [[claim-ai-productivity-enabler]] and directly raises [[open-question-ai-data-privacy]]. Attributed to [[entity-indra-nooyi]].

**Enrichment.** Strongly supported by independent LLM evaluations documenting hallucination, shallow synthesis, and training-cutoff 'currentness' limits without retrieval augmentation; academic critiques agree GenAI gives useful 'first drafts' but needs expert supervision for nuanced decisions. Counter-trend: domain-specific fine-tuning and retrieval-augmented generation (RAG) are steadily narrowing these gaps for routine tasks.


## Related across articles
- [[concept-judgment-debt]]
- [[claim-human-capital-roi]]