---
id: "claim-public-llms-low-value"
type: "claim"
source_title: "Don't Let AI Slop Muck Up Your Company's Processes"
source_url: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
source_timestamps: ["§ 3. Define what value is being added."]
tags: ["technology-strategy", "model-selection"]
related: ["action-use-proprietary-slms"]
speakers: ["Matthias Holweg", "Thomas H. Davenport"]
confidence: "medium"
testable: true
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-sig-54-ai-slop-processes"
sourceUrl: "https://hbr.org/2026/06/dont-let-ai-slop-muck-up-your-companys-processes"
sourceTitle: "Don’t Let AI Slop Muck Up Your Company’s Processes"
---
# Public LLMs add little value compared to proprietary SLMs

**Claim:** For many business tasks, public LLMs add little to no real value because they merely create generic prose prone to mistakes. True value comes from proprietary Small Language Models (SLMs) or larger models customized on proprietary data to generate insights, while public models such as [[entity-chatgpt-d54|ChatGPT]] and [[entity-claude-d8|Claude]] are used merely as formatting or styling engines.

This is operationalized in [[action-use-proprietary-slms]].

**Confidence:** medium (author) / *strategic emphasis is well aligned with governance guidance, but the blanket statement is overstated and not empirically established* (enrichment). The push toward proprietary, domain-tuned models matches NIST's data-provenance and objective-clarity guidance and industry practice for regulated/high-stakes use. But no cited comparative study proves public LLMs 'add little value'; many documented use cases (coding copilots, drafting assistants) show substantial value from public models. Read this as strategic positioning around competitive differentiation via proprietary data. **Testable:** yes.


## Related across articles
- [[claim-proprietary-models-not-competitive-advantage]]
- [[contrarian-off-the-shelf-over-proprietary]]
- [[action-use-proprietary-slms]]