After attending Adobe Summit this week, one theme stood out: AI isn’t just about capabilities and features—it’s about where it replaces vs. augments human work, and what that does to teams and outcomes. So, for This Week in AI for April 21, that’s where we’re focusing.
AI can impact team collaboration
While AI can boost productivity and insight, it can come at the expense of long-term effectiveness and teaming. When AI is used to augment collaboration and judgment, it’s a force multiplier. But things start to break down when AI replaces instead of enhances: reduced persistence, weaker collaboration, blurred accountability. A divide is forming between teams that use AI to think better—and teams that use it to think less.
- In one study by a Duke University professor, class teams jumped from ~5% to over 50% after AI tools were introduced, as individuals opted to work alone rather than together (Fast Company).
- Coinbase is piloting AI agents that participate in Slack and email like human coworkers or mentors, raising new questions around trust, ownership, and decision-making authority (The Block).
- A recent research paper found that while AI improves immediate performance, users perform worse and give up more quickly once it’s removed—even after just ~10 minutes of use (arXiv 2604.04721).
- AI’s “jagged intelligence” means it replaces tasks, not whole jobs. It excels in some areas and fails in others, making human integration and oversight critical (The New York Times).
- That’s why augmentation is emerging as the stronger strategy—companies that enhance human work outperform those that try to replace it (Harvard Business Review).
- But even when leaders know this, competition can push companies into over-automation anyway, creating a race to the bottom (arXiv, “The AI Layoff Trap”).
- Research from BetterUp shows “AI power users” are significantly more productive and engaged, while others fall behind—suggesting AI may widen performance gaps within teams (BetterUp).
At Adobe Summit, a consistent theme was using AI to enhance, not replace, human decision-making. As roles evolve, the goal isn’t to remove designers and marketers, but to give them better opportunity to experiment, iterate, and use their judgment.
“Claude Design” isn’t killing design—it’s commoditizing it
AI is making it easier for anyone to produce work that looks, functions, and appears “good enough,” but most of that work will default to beige. As tools like Anthropic’s Claude Design and Canva AI make design execution more accessible, the role of designers isn’t disappearing, but it is changing. The real value is no longer just execution: designers need to push beyond what AI would generate on its own.
If AI raises the floor, we have to raise the ceiling, too.
- Claude Design frames AI as a system that can generate interfaces and experiences from prompting, signaling a move toward “design via conversation” rather than traditional tools (Anthropic). Canva is also doubling down on this shift, positioning AI as a way to let anyone create presentations, visuals, and content instantly. This continues their mission to democratizing design and make design/creative tools more accessible (Canva Create 2026).
- Adobe is taking a different angle: embedding AI into enterprise workflows as “coworkers” that assist across the customer experience—from content generation to orchestration—rather than replacing creative professionals outright (Adobe).
This introduces new risks: non-experts may overestimate the quality of AI-generated outputs, leading to shallow experiences that just become noise. Without strong leadership, organizations risk producing more and accomplishing less.