Validate. Verify. Authenticate.

Your critical thinking skills are extremely important in the age of AI.

Why It Matters: Your ability to employ critical thinking protects you from any crazy AI answers or nefarious deep fakes. It also makes you a better user of AI tools.

The words form a mantra in your brain that is easily memorable: Validate. Verify. Authenticate.

The words seem similar. But I like using all three because they bring their own separate emphasis.

Validate: This examines if something meets standards or makes sense in context. It assesses soundness. Assesses reasonableness. Or even appropriateness. It is the big umbrella. Does this "thing" even in the ballpark of what is sound?

Verify: Dives a bit deeper. Determines factual accuracy or truth. Checks statements and data against other evidence or means of independent confirmation.

Authenticate: focuses on confirming the identity of a source. Proving whether something is what it claims to be or comes from where it claims to come from.

These can be applied to two different modes of AI.

GenAI has two primary modes of interaction:

Research Mode for retrieving and synthesizing external information, and Reasoning Mode for generative thinking and ideation.

More specifically:

• Research Mode - The AI retrieves and synthesizes information from external sources (web search, databases, documents). It acts as an intelligent research assistant finding existing information.
• Reasoning Mode - The AI generates responses based on its training, serving as a thought partner for brainstorming, analysis, and creative problem-solving without needing to access external sources.

Research Mode (Retrieval-Augmented)

How each applies:

• Validate: Does the retrieved information make sense for my question? Are these sources relevant and reasonable?
• Verify: Are the facts and claims in these sources accurate? Can I cross-reference them?
• Authenticate: Are these sources who they claim to be? Is this really from PMI, a certified expert, or an established methodology?

Example: You ask "What are the current best practices for Agile sprint planning?"

• Validate: The AI retrieves Agile methodology guides (makes sense) rather than waterfall documentation (doesn't make sense for this question)
• Verify: Cross-checks the sprint planning steps across multiple sources (Scrum Guide, PMI resources, Atlassian documentation)
• Authenticate: Confirms information comes from official Scrum.org or PMI.org, not from "agile-tips-blog.net"

Reasoning Mode (Generative/Knowledge-Based)

How each applies:

• Validate: Does the AI's reasoning follow logical principles? Is the advice contextually appropriate?
• Verify: Can I test the AI's claims against my own knowledge or other sources? Do the logical steps hold up?
• Authenticate: This is trickier—you're verifying the AI is drawing from legitimate training rather than hallucinating or confabulating

Example: You ask "How should I handle a team member who consistently misses deadlines?"

• Validate: The advice follows sound management principles (one-on-one conversation, identifying root causes) and fits your organizational context
• Verify: The suggested approach (document issues, create improvement plan) aligns with established performance management frameworks you know
• Authenticate: The AI isn't citing fake PM experts or attributing methodology to organizations that didn't create it (like claiming "The Jones Method" exists when it doesn't)

Validate. Verify. Authenticate.

Three solid definitions to check yourself when using GenAI.

These are foundational and important for all project managers.

The very last thing you want to do with stakeholders is submit project use cases that were hallucinated and not real.