Skip to main content

Why AI in marketing does not equal working AI

Why AI in marketing does not equal working AI

The gap between marketing AI and working AI is the central problem DecAIHub addresses. Hundreds of crypto projects use words like "AI-powered," "built with cutting-edge AI," or "our AI engine" in their branding — but when you look for verifiable artifacts, many have no code, no model, no benchmarks, and no technical documentation. This page explains why marketing language alone is not evidence, how to identify the most common patterns of exaggeration, and how to tell the difference using a DecAIHub project card.

In short: anyone can write "AI" on a website. What matters is whether a working AI component exists behind the claim — and whether Tier-1 evidence supports it.

Common marketing patterns

Certain phrases appear so frequently across crypto project websites that they have become reliable indicators of positioning without substance:

  • "AI-powered" or "powered by artificial intelligence" — used as a modifier with no technical context.
  • "Our proprietary AI engine" — no architecture description, no model name, no version history.
  • "Built with cutting-edge AI/ML" — no specification of what kind of ML, what data, or what output.
  • "AI" in the token name or ticker — the word appears in branding but not in any technical artifact.
  • "AI features coming soon" or "AI roadmap Q3" — intent without implementation.
  • "Trained on billions of data points" — no dataset description, no methodology, no verifiable source.
  • "Advanced neural network" — no architecture type, no layer count, no training details, no benchmarks.

These phrases are not automatically false. But without supporting artifacts, they carry zero verifiable weight. The key test is whether the language is accompanied by specifics: a named model architecture, a published dataset, a linked repository, or a working demo.

How marketing claims escalate

Marketing AI claims tend to follow a predictable escalation ladder. Understanding this pattern helps distinguish genuine communication from positioning:

Stage Typical claim What is missing
Name-level "AI" in token name or project title Any description of what the AI does
Modifier-level "AI-powered platform" on landing page Any technical detail or architecture
Feature-level "Our AI analyzes market data in real time" Model type, data source, accuracy, demo
Architecture-level "Built on a proprietary transformer model" Repository, version, benchmark, training data
Evidence-level "See our model card, benchmark results, and API docs" Nothing — this is what verification looks like

Most projects that trigger AI-washing concerns are stuck at the Name or Modifier stage. Projects at the Evidence level almost always have verifiable AI components.

Why marketing alone is not evidence

Writing "AI" in a whitepaper, landing page, or pitch deck requires no technical capability. There is no industry standard that prevents a project with no AI from calling itself AI-powered. No regulatory body audits these claims in real time. This creates an environment where the cost of claiming AI is near zero, while the cost of building real AI is substantial.

The result is a predictable gap: projects with heavy AI branding may have ai_score = 0 in DecAIHub (no verifiable AI found), while projects with minimal marketing but real technical work score ai_score = 3. The label tells you what the project wants to be perceived as. The evidence tells you what it demonstrably is.

This asymmetry is not unique to crypto. In traditional technology sectors, companies regularly face scrutiny for overstating AI capabilities. What makes the crypto AI space different is the additional layer of tokenization: a project's narrative directly affects token perception, which amplifies the incentive to exaggerate technical claims.

The verification checklist

When reading a project card, check these areas before trusting an AI claim:

  1. AI component field — is there a specific, descriptive value (e.g., inference engine for LLM routing, agent framework + SDK, GPU orchestration), or does it say positioning only, none, or unknown?
  2. ai_score — a score of 0 means no Tier-1 evidence of working AI was found. A score of 3 means strong, verifiable artifacts exist.
  3. Evidence tier — are there public repositories with AI code, technical documentation with versions and dates, or working demos?
  4. Specificity — does the project name a concrete model, architecture, dataset, or benchmark? Vague language is a warning sign.
  5. Timeline — was "AI" always part of the project, or was it added to the name and positioning after launch with no corresponding technical change?
  6. Consistency across sources — do the whitepaper, GitHub repository, official documentation, and website tell the same technical story? Divergence between sources often indicates that marketing has outpaced engineering.

A practical approach: start with the AI component field and ai_score. If the component says positioning only and the score is 0, the remaining checks are unlikely to reveal hidden substance. If the component names a specific function and the score is 2 or higher, the deeper checks help you understand how strong the evidence is.

See How to tell if a project really has AI and What counts as evidence of AI for a detailed walkthrough.

How DecAIHub handles this: AI_0 vs AI_1 through AI_4

The DecAI Fit metric separates intent from evidence using granular flags:

  • AI_0 — the project has a Tier-1 declaration of AI intent (roadmap, official docs), but no evidence of a working AI component. This is the "marketing only" baseline.
  • AI_1 — the project describes what exactly the AI does (specific function, not just a buzzword).
  • AI_2 — the project has working demos or examples where the AI component produces observable output.
  • AI_3 — the project publishes metrics, benchmarks, or quality assessments with methodology.
  • AI_4 — the project provides technical documentation: architecture, model versions, data pipeline descriptions.

A project stuck at AI_0 has declared intent but provided no verifiable substance. The distance between AI_0 and AI_1+ is exactly the gap between marketing AI and working AI.

Each flag is binary (0 or 1) and independently assessed. A project can have AI_1 = 1 and AI_3 = 0, meaning it describes what the AI does but has not published quality metrics. The flags build a profile, not a single score — and that profile reveals precisely where evidence stops and marketing begins.

The gap in practice

The pattern repeats across the DecAIHub registry:

  • Heavy AI marketing, low ai_score: the project's website, social accounts, and pitch materials prominently feature "AI" — but the AI component field reads positioning only and the ai_score is 0. No code repository, no model card, no benchmark, no demo.
  • Minimal marketing, high ai_score: the project barely mentions AI in its branding — but the card lists a specific AI component, links to a repository with model code, references published benchmarks, and documents a versioned architecture. The ai_score is 3.

The first project is louder. The second project is more verifiable. DecAIHub measures the second dimension.

When rebranding signals AI-washing

One of the most recognizable patterns in the crypto AI space is post-launch AI rebranding: a project that launched without any AI narrative adds "AI" to its name, ticker, or positioning after the concept gains attention. This is not an accusation against the project, but it is a strong signal worth investigating.

Key questions to ask when a project rebrands to AI:

  • Was there a corresponding technical change? If the project added AI to its name but no new code, model, or technical documentation appeared around the same time, the rebranding is cosmetic.
  • Does the AI component field reflect the change? A genuine AI pivot would result in a specific AI component description. If the field still reads positioning only, the rebrand has not been substantiated.
  • Is there a timeline of development? Real AI integration leaves traces: repository commits, documentation updates, model version changes. A rebrand without these traces is positioning, not engineering.

Connection to AI-washing

When the gap between marketing claims and verifiable evidence is wide, DecAIHub calls this AI-washing — presenting a project as AI-driven when no supporting artifacts exist. AI-washing is not an accusatory verdict; it is a measurable signal: bold AI claims paired with positioning only or ai_score = 0.

AI-washing is especially common in categories where "AI" adds perceived value to a token or narrative, such as meme tokens with AI narrative or projects that rebrand as AI after launch.

What this concept alone does not prove

  • Heavy marketing does not prove the project has no AI. Some projects with strong marketing also have real, documented AI components.
  • A low ai_score is not an accusation against the project. It means Tier-1 evidence of working AI was not found as of the verification date. See Why absence of evidence is not absence of technology.
  • Reaching AI_1+ does not prove the AI is high quality. It means verifiable artifacts exist, not that they demonstrate excellence.
  • A high ai_score does not prove the project solves a real problem. It proves that AI artifacts are documented, not that they are useful.

Frequently asked questions

How common is AI-washing in crypto?

AI-washing is widespread across the crypto AI space. A significant portion of projects that include "AI" in their branding or token name show no verifiable AI component when assessed against Tier-1 evidence criteria. DecAIHub tracks this systematically: any project with ai_score = 0 and prominent AI positioning is flagged as a potential AI-washing case. The pattern is especially prevalent in categories where narrative drives attention, such as meme tokens or newly launched projects that adopt AI branding.

Can a project with heavy marketing still have real AI?

Yes. Strong marketing and genuine AI are not mutually exclusive. Some well-funded projects actively promote their AI capabilities and also maintain public repositories, published benchmarks, and detailed technical documentation. The test is not whether marketing exists, but whether it is backed by verifiable evidence. In DecAIHub terms, a project with aggressive marketing and ai_score = 3 has passed the verification threshold regardless of how loudly it promotes itself.

What is the fastest way to check if AI claims are real?

Start with two fields on the project card: the AI component field and the ai_score. If the AI component reads positioning only or none and the ai_score is 0, the project has no verified AI regardless of its marketing. If the AI component names a specific function and the ai_score is 2 or higher, proceed to check the evidence tier and linked documentation for depth.

Does rebranding to AI after launch indicate AI-washing?

Not automatically, but it is a strong warning sign. Legitimate AI pivots happen when a project genuinely integrates AI technology — which leaves traces in code repositories, documentation, and technical announcements. If the only change is a new name or updated website copy with no corresponding technical artifacts, the rebranding is cosmetic and likely qualifies as AI-washing in DecAIHub's framework.

External references

See also

No comments to display

Back to top