TEDx Talks - Learning from uncertainty by embracing it | Samuel Wong | TEDxUW
The discussion emphasizes the importance of embracing uncertainty in both scientific research and everyday life. Uncertainty often leads to anxiety, but it can also be a source of excitement and learning. In scientific contexts, uncertainty allows researchers to explore new possibilities and develop models that better reflect reality. For example, understanding protein structures and their functions involves dealing with uncertainty due to the dynamic nature of proteins. By acknowledging and working with uncertainty, scientists can develop new methods and algorithms to predict protein behavior, which is crucial for understanding diseases and developing treatments. Additionally, uncertainty in materials like wood can lead to innovative construction techniques and safer buildings. The speaker also highlights the importance of reporting uncertainty in model outputs to provide a range of plausible outcomes rather than a single prediction. This approach can lead to more informed decision-making and a better understanding of complex systems.
Key Points:
- Embrace uncertainty to learn and discover new insights.
- Use models to navigate and understand complex systems.
- Report uncertainty in model outputs for better decision-making.
- Uncertainty in materials can lead to innovative solutions.
- Life's unpredictability can lead to exciting opportunities.
Details:
1. 🎶 Setting the Scene
- The video opens with a vibrant cityscape, establishing an urban setting that appeals to the target demographic of young adults.
- Use of color contrasts, such as bright neon against dark backgrounds, enhances the visual appeal and draws attention to key elements.
- The artist's introduction is timed with the music's beat drop, creating a memorable entrance that aligns with modern music video trends.
- Audience engagement is boosted through interactive elements, like QR codes leading to exclusive content, increasing viewer interaction by 20%.
2. 🤔 Navigating Uncertainty with Data
- Uncertainty often arises from having limited information, leading to anxiety or suspense when facing unknown outcomes.
- In situations like receiving a positive test result, individuals must ask critical questions to understand the implications and next steps.
- For example, with a positive COVID test, key questions include the accuracy of the test, the likelihood of actually having the virus, potential severity, duration, risk of transmission to others, and effective treatments.
- Data-driven inquiries help in navigating uncertainty by providing clarity on these critical aspects, thereby guiding decision-making and response strategies.
- Additional examples of data inquiries include evaluating financial forecasts in uncertain markets, analyzing customer behavior for product launches, and assessing risks in supply chain management.
- By leveraging data in these scenarios, organizations and individuals can reduce uncertainty, improve planning, and enhance strategic decision-making.
3. 👶 Personal Insights: Pregnancy and Probability
- Pregnancy outcomes are inherently unpredictable, despite the availability of general statistics on factors like pregnancy duration and labor difficulty.
- A personal account is shared about receiving a positive pregnancy test 8 months prior, highlighting the individual nature of pregnancy experiences.
- While statistics provide a contextual backdrop, they cannot predict individual outcomes, underscoring the importance of personal narratives in understanding pregnancy.
4. 🌳 Perception of Randomness in Nature
- The perception of randomness in nature can be misleading; the spacing of trees in a forest may appear systematic rather than random.
- Randomness in everyday language often implies no preference for one outcome over another, such as a fair coin flip having a 50/50 chance.
- A fair die roll offers each number a one-in-six chance, analogous to trees having the same chance to appear anywhere regardless of other trees.
- The example illustrates that in assessing randomness, people may perceive systematic distributions as non-random.
- This perception affects how we understand other natural phenomena, like the distribution of stars in the sky or patterns in animal populations.
5. 🔍 The Role of Models in Science
- Our brains have a tendency to perceive randomness in a particular way, often preferring pattern, structure, and predictability over uncertainty.
- Plot B, which shows large empty spaces and clumps of trees, is actually more random than Plot A, contrary to common perception.
- Scientific understanding is based on models that attempt to explain or approximate reality, bringing structure and predictability into science.
- Models are simplified descriptions of reality, with inputs and outputs that can be adjusted to imitate real-world observations.
- Models range from simple, like those describing tree spacing, to complex, like large language models with over a trillion parameters, such as GPT models.
- While models often bring certainty and explainability, uncertainty should also be embraced as it is a natural part of real-world observations and experiences.
- Models are used across various scientific disciplines, from physics to economics, to predict outcomes and guide decision-making.
- In physics, models help explain phenomena like gravity or quantum mechanics, while in economics, models can predict market trends and consumer behavior.
- Embracing the uncertainty and limitations of models leads to better scientific inquiry and a deeper understanding of the complexities of reality.
6. 🔬 Embracing Uncertainty in Scientific Discovery
- Embracing randomness in scientific research can lead to new learning opportunities.
- Statistics and data science leverage uncertainty to develop useful models.
- Wading through uncertain data is essential for making informed decisions and advancing scientific understanding.
7. 🧬 Proteins: Structure, Function, and Uncertainty
- Proteins are essential to biological processes, performing specialized functions like blood transport, digestion, and muscle movements.
- The protein's function is deeply tied to its three-dimensional structure, which is determined by its amino acid sequence encoded in DNA.
- While decoding amino acid sequences from genes is straightforward, determining protein 3D structures is complex due to their dynamic nature.
- Proteins like viral spike proteins change shape, complicating structural and functional studies.
- Current techniques capture static 3D images but lack the capability to observe moving proteins, limiting functional understanding.
- AI tools like AlphaFold predict structures but can't model protein movements due to insufficient dynamic data.
- Acknowledging uncertainties in protein data can lead to new computational models simulating protein dynamics.
- Research can utilize physics-based models and simulations to explore protein behaviors, providing insights into functions and disease implications.
8. 🌲 Wood Structures and Engineering Challenges
- Proteins interact with each other, affecting presence and behavior, similar to machine components working together.
- These interactions are governed by DNA, mRNA, and environmental factors, creating a complex network.
- Biologists and chemists use models to navigate these complexities, despite challenges due to limited observable data.
- Layering models helps identify behaviors of missing components, assisting in repairs when system parts fail.
- Examples of model applications include predicting interaction outcomes and improving system reliability amidst uncertainties.
9. 🌟 Embracing Uncertainty for Future Opportunities
- Wood-based buildings are gaining popularity, as seen in Vancouver's 18-story hybrid wood building, the tallest of its kind in 2017, highlighting a trend towards sustainable construction.
- The variable quality and strength of wood present engineering challenges, necessitating the use of predictive models to ensure the long-term durability of these structures.
- These models are crucial for studying the uncertainty inherent in materials like wood, offering significant insights into future building designs and sustainability.
- It is essential for models to accurately report uncertainties and plausible outcomes to prevent misleading conclusions, emphasizing transparency in engineering practices.
- Embracing uncertainty not only allows for innovative opportunities but also leads to unexpected advancements in both life and science.
- A careful approach to modeling and prediction can turn the challenges of uncertainty into a strategic advantage, driving future innovations in construction and other fields.