In Episode 9 of the Intentional AI series, Cole and Virgil take on one of the most common and misunderstood uses of AI today: image and graphic generation. From social media visuals to promotional graphics, AI images are fast, easy, and everywhere.

The conversation focuses on why images became the public on ramp to AI and why that familiarity creates risk. Visuals feel harmless, but the moment AI starts generating finished looking images, teams inherit decisions around ownership, ethics, and trust that they are often unprepared to make.

A central theme of the episode is responsibility escalation. As AI reduces the effort required to create images, the importance of human judgment increases. Treating AI generated visuals as final work can quickly introduce legal, ethical, and reputational problems.

Virgil shares a practical experiment where he used a simple prompt to generate three social media promotional graphics from an existing article and tested the results across three tools: Canva, Claude, and Artlist.

Canva produced the most generic and repetitive designs. Claude delivered cleaner structure and stronger messaging but struggled with fonts, formats, and variation. Artlist created the most visually interesting outputs, though it introduced workflow limitations and cost concerns.

The episode reinforces a consistent conclusion across the series. AI can help jumpstart visual work, but it cannot replace judgment, intent, or responsibility.

In this episode, they explore:

1. Why AI images are so tempting to use
2. Where AI generated graphics actually help
3. Why most AI visuals fall flat
4. Ethical and ownership risks teams overlook
5. A comparison of Canva, Claude, and Artlist

Previously in the Intentional AI series:

1. Episode 1: Intentional AI and the Content Lifecycle
2. Episode 2: Maximizing AI for Research and Analysis
3. Episode 3: Smarter Content Creation with AI
4. Episode 4: The role of AI in content management
5. Episode 5: How much can you trust AI for accessibility
6. Episode 6: You’re asking AI to solve the wrong problems for SEO, GEO, and AEO
7. Episode 7: Why AI can make your content personalization worse
8. Episode 8: The real value of AI wireframes is NOT the wireframes

New episodes every other Tuesday.

For more conversations about AI, design, and digital strategy, visit www.discussingstupid.com and subscribe on your favorite podcast platform.

(0:00) - Intro

(1:40) - You can’t escape AI imagery

(3:18) - Why AI images are risky

(4:40) - The legal and ethical line

(6:15) - Creativity vs time and cost

(9:28) - Every tool has hopped on the AI bandwagon

(13:20) - The slippery slope of AI visuals

(15:35) - We tested 3 tools for AI visuals

(17:30) - Testing Canva

(20:40) - Testing Claude...

In Episode 8 of the Intentional AI series, Cole, Virgil, and Chad explore one of the most tempting uses of AI in digital work: wireframing and page layout. With AI now able to generate full wireframes in minutes or even seconds, the promise of speed is undeniable. But speed alone is not the point.

The conversation focuses on where AI genuinely helps in the wireframing process and where it introduces new risks. Wireframes are meant to establish structure, hierarchy, and intent, not just visual output. While AI can quickly generate layouts, components, and patterns, it still requires strong human judgment to evaluate what is correct, what is missing, and what could cause problems downstream.

A key theme of the episode is escalation of responsibility. As AI reduces the time required to create wireframes, the importance of human review, direction, and decision making increases. Treating AI generated wireframes as finished work can introduce serious risks, especially around accessibility, content fidelity, maintainability, and overall project direction.

Virgil shares an experiment where he used AI to first generate a detailed prompt for wireframing, then tested that prompt across three tools: Claude, Google Gemini 3, and Figma Make. The results reveal clear differences in layout quality, accessibility handling, content retention, and how easily the outputs could be integrated into real workflows.

Claude produced the strongest layout and structural patterns but failed badly on accessibility and removed large portions of content. Gemini generated simpler wireframes with clearer structure, but used even less content and still struggled with accessibility. Figma Make stood out for workflow integration, retaining all content and allowing direct editing inside Figma, though it also failed accessibility requirements and relied heavily on generic styling and placeholder imagery.

Throughout the episode, the group returns to the same conclusion. AI is extremely effective at getting the first portion of wireframing done quickly. It is far less effective at making judgment calls, enforcing standards, or understanding context without guidance.

In this episode, they explore:

1. How wireframing fits into the content lifecycle
2. Why speed changes the risk profile of design work
3. Using AI to generate prompts instead of starting from scratch
4. Where AI wireframes succeed and where they fail
5. Accessibility and content risks in AI generated layouts
6. A wireframing comparison of Claude, Gemini 3, and Figma Make

A downloadable Episode Companion Guide is available below with tool comparisons and key takeaways.

DS-S3-E8-CompanionDoc.pdf

Previously in the Intentional AI series:

1. Episode 1: Intentional AI and the Content Lifecycle
2. Episode 2: Maximizing AI for Research & Analysis
3. Episode 3: Smarter Content Creation with AI
4. Episode 4: The role of AI in content management
5. Episode 5: How much can you trust AI for...

In Episode 7 of the Intentional AI series, Cole and Virgil focus on content personalization and why it is one of the most overpromised areas of AI. While personalization is often positioned as simple and automated, doing it well requires far more clarity and intent than most tools suggest.

They break personalization into two main approaches. Role based personalization tailors messages for specific audiences or job functions, while behavioral personalization adapts experiences based on how people interact with content over time. The conversation also touches on predictive analysis and where AI may eventually help interpret patterns across analytics data.

A central theme of the episode is trust. Using AI for personalization assumes the system understands audience priorities and pain points. Without clear direction, AI fills in the gaps with assumptions. Cole and Virgil explain why personalization has always been difficult to implement, why adoption remains low, and why AI does not remove the need for strategy, measurement, or human judgment.

The episode also addresses the risks of personalization. Messages that are too generic get ignored, while messages that feel overly personal can cross into uncomfortable territory. Finding the right balance is still a human responsibility.

In the second half of the episode, they continue their ongoing experiment using the same AI written accessibility article from earlier episodes. This time, they test three tools by asking them to generate role based promotional emails for a head of web marketing, a director of information technology, and a C level executive. The results highlight meaningful differences in tone, structure, and assumptions across tools.

The takeaway is consistent with the Intentional AI series. AI can support personalization, but only when you define goals, outcomes, and boundaries first.

In this episode, they explore:

1. What content personalization actually means
2. Role based versus behavioral personalization
3. Why personalization adoption remains low
4. The balance between relevance and creepiness
5. How AI supports personalization without replacing strategy
6. A role based email comparison of Perplexity, Copilot, and Claude

A downloadable Episode Companion Guide is available below with tool comparisons and practical takeaways.

DS-S3-E7-CompanionDoc.pdf

Previously in the Intentional AI series:

1. Episode 1: Intentional AI and the Content Lifecycle
2. Episode 2: Using AI for Research and Analysis
3. Episode 3: AI and Content Creation
4. Episode 4: Content Management and AI
5. Episode 5: How much can you trust AI for accessibility?
6. Episode 6: You’re asking AI to solve the wrong problems for SEO, GEO, and AEO

New episodes every other Tuesday.

For more conversations about AI and digital strategy, visit