# AI 4 Good | Krzysztof Wojewodzic | TEDxKoszalin

Source: https://www.youtube.com/watch?v=uOrowTdDBWs
Recap page: https://rapidrecap.app/video/uOrowTdDBWs
Generated: 2026-01-14T16:09:18.888+00:00

---
## Quick Overview

Krzysztof Wojewodzic presents how Artificial Intelligence (AI) for Good, specifically through tools like PreferenceOS, can significantly optimize critical operational areas in healthcare, such as emergency response times and administrative tasks, ultimately saving time and potentially lives by allowing professionals to focus on what truly matters.

**Key Points:**
- AI optimization in emergency medical services (EMS) reduced the time needed to transfer critically ill patients to the hospital by 1.5 hours, as illustrated by a case involving a heart patient requiring blood transfusions.
- AI tools can predict the duration of medical procedures, improving scheduling accuracy; for example, predicting surgery times that were previously overestimated by 3 hours.
- AI is being utilized to analyze and summarize medical documentation, which currently consumes 30-50% of a doctor's time during home visits.
- The presentation highlights the application of AI in understanding human behavior, showing models that detect emotions (60% accuracy) and attention (80% accuracy) using techniques like YOLO and gaze estimation.
- The speaker challenges the audience to consider the ethical direction of AI, asking whether it will be used for trivialities like generating 'Instagram models' or for solving real-world problems (e.g., optimizing streaming choices or improving healthcare).
- The speaker shares his personal AI toolkit, available via a QR code, encompassing methods like heuristic optimization, predictive models, and reinforcement learning (RL) for scheduling.
- The core message is that the future benefit of the AI revolution depends on how society chooses to deploy these powerful tools.

![Screenshot at 00:22: A yellow and red emergency rescue helicopter \(Lotnicze Pogotowie Ratunkowe\) is shown landing or taking off, visually representing the critical, life-saving applications of technology in emergency medical response discussed in the presentation.](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-00-22.jpg)

**Context:** This is a TEDx talk given by Krzysztof Wojewodzic at TEDxKoszalin, focusing on the concept of "AI 4 Good"—applying artificial intelligence for positive societal impact. The speaker contrasts trivial or superficial uses of AI (like creating fake news or optimizing entertainment choices) with critical applications in healthcare and operational efficiency, using real-world examples from emergency services and medical administration.

## Detailed Analysis

Krzysztof Wojewodzic argues for directing AI development towards beneficial applications ("AI 4 Good"). He illustrates the impact of AI in emergency medical services (EMS), citing an example where AI optimization reduced the time required to transport a critically ill patient needing blood transfusions to the hospital by 1.5 hours. Furthermore, he discusses how AI can predict surgery times to improve hospital scheduling, potentially saving up to 12% more operations, and automate the 30-50% of a doctor's time spent on administrative documentation. Wojewodzic then pivots to AI in understanding human behavior, presenting models capable of detecting emotions with 60% accuracy and attention with 80% accuracy, using techniques like YOLO and gaze estimation. He contrasts these serious applications with frivolous uses, such as generating TikTok-style content or creating misleading images ('deepfakes' of politicians or sensationalized Mona Lisas). He concludes by emphasizing that the direction of the AI revolution—whether it leads to better health and learning or merely reinforces superficial consumption—is entirely up to collective human choices, urging the audience to utilize AI responsibly. He provides a link to his personal AI toolkit for others to explore these tools.

### Emergency Medical Services (EMS) Optimization

- AI reduced patient transfer time by 1.5 hours in a critical case involving a heart patient needing blood
- AI predicts procedure times to optimize scheduling, potentially increasing operations by 12%

### Healthcare Administration Improvement

- AI is used to summarize medical documentation, reclaiming 30-50% of doctor time currently spent on paperwork during home visits

### AI for Human Understanding

- Models trained using YOLO and gaze estimation can detect emotions (60% accuracy) and attention (80% accuracy) in individuals

### The Dangers of Misuse

- Examples shown include AI-generated 'fake news,' misleading images (e.g., sexualized Mona Lisa), and the 'choice paralysis' caused by algorithms on streaming platforms

### Methods and Tools

- The speaker's approach involves heuristics/optimization, predictive models, knowledge graphs, and affective cognitive maps
- He shares his 'AI Toolkit' via a QR code for audience access

### Conclusion and Responsibility

- The ultimate direction of the AI revolution depends on human choices—whether to use it for trivial distractions or for solving fundamental problems like health and education

![Screenshot at 00:22: A yellow and red emergency rescue helicopter \(Lotnicze Pogotowie Ratunkowe\) is shown landing or taking off, visually representing the critical, life-saving applications of technology in emergency medical response discussed in the presentation.](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-00-22.jpg)
![Screenshot at 01:37: A slide illustrating the problem of administrative load in healthcare, showing a doctor using a laptop while writing on a clipboard, with text noting that 30-50% of visits are consumed by documentation.](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-01-37.jpg)
![Screenshot at 02:51: A slide showcasing examples of problematic AI outputs \(memes, deepfakes\) next to the text 'FAKE NEWS' and 'DEEP FAKE', used to illustrate the misuse of generative AI.](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-02-51.jpg)
![Screenshot at 05:59: A slide detailing the current effectiveness of AI in emotion detection \(60%\) and attention detection \(80%\), contrasting it with the need for better performance.](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-05-59.jpg)
![Screenshot at 11:11: The final slide showing the speaker, Dr. Krzysztof Wojewodzic, with a QR code linking to his 'MÓJ AI TOOLKIT' and the title 'RZEMIEŚLNICZA PREZENTACJA' \(Craft Presentation\).](https://ss.rapidrecap.app/screens/uOrowTdDBWs/00-11-11.jpg)