---
id: "claim-worst-ai-today"
type: "claim"
source_timestamps: ["§ Systems Thinking"]
tags: ["technology-evolution", "future-of-work"]
related: ["quote-worst-ai", "concept-systems-thinking-ai"]
confidence: "high"
testable: false
speakers: ["Tom Davenport", "John J. Sviokla"]
sources: ["spine"]
sourceVaultSlug: "hbr-seg-spine"
originDay: 1
articleStem: "hbr-cl-95-6-disciplines-genai"
sourceUrl: "https://hbr.org/2024/07/the-6-disciplines-companies-need-to-get-the-most-out-of-gen-ai"
sourceTitle: "The 6 Disciplines Companies Need to Get the Most Out of Gen AI"
---
# Current AI is the worst we will ever use

**Claim:** Because the technology is evolving so rapidly (only **18 months** into reinventing generative work at the time of writing), the generative AI models we use today are the **least capable versions we will ever have**. The verbatim line is [[quote-worst-ai]]; the urgency it creates motivates acting on [[concept-systems-thinking-ai]] now.

**Confidence: high · Testable: no** — this is primarily a *rhetorical and forward-looking* claim about the expectation of rapid, continuous improvement, not an empirically testable statement.

Enrichment validation (trend-based): the trajectory GPT-3 → GPT-3.5 → GPT-4, with gains in accuracy, reasoning, and multimodality, supports the assertion of fast recent improvement; scaling laws, new architectures, and multimodal integration suggest continued short-term progress.

**Limitations / counter-perspectives:** progress may plateau or face regulatory, safety, energy, or hardware constraints; it is not guaranteed every future model is superior on *all* dimensions (some may trade open-endedness for safety). Some experts argue we are entering diminishing returns on pure scaling, with slower or more domain-specific future gains.