PROJECTS UNDER MENTORING
Cohort: April - July 2024
- All Projects
- Artificial Inteligence
- Biosecurity/Biosafety
- Global Causes
- Communication
AI Capabilities in Graph Problem Solving
This project, wilI intend to propose graph problem-solving tasks for different models, taking the BIG-benchmark format as a starting point and using models such as BLOOM, GPT-3, GPT-4, T5, FLAN-T5, and BERT.
[…][…]
Regulating Change: The Urgency of Standards for the Responsible Advancement of AI
This project investigates the growth trends in Artificial Intelligence (AI) models, focusing on the processing capacity and energy consumption of data centers worldwide. The project will analyze the factors influencing energy consumption, evaluate various data centers, and propose optimizations to improve energy efficiency without compromising performance. It will include a literature review, data analysis, predictions, risk assessments, and ultimately, the drafting of a report detailing the findings. This report will serve as an information source for responsible and sustainable management of AI in the future.
[…][…]
Open-source Replication of Decomposition of Language Models with Dictionary Learning
This project will replicate Anthropic’s paper on Dictionary Learning for feature extraction from Language Models [Bri+23], and explore potential extensions if possible.
⭐ Fernando was accepted to participate on the ML4Good Bootcamp during his mentorship in Carreras con Impacto.
[…][…]
Competence in Scalability and Performance of AI Models Based on Benchmarks Between the United States and China
This aims to conduct a comprehensive and comparative analysis to inform and raise public awareness about the risks associated with Artificial Intelligence (AI) and the necessity of specific regulations. Utilizing a matrix of factors, the project will compare AI regulations across Latin American countries and the European Union, which has recently passed the AI Act. This matrix will explore key areas such as existing AI laws, investment levels in AI research, technological advancement, data protection policies, and leading national AI-using companies. The project seeks to highlight the importance of fairness and justice in AI algorithm design, data privacy, security, and the authenticity of AI-generated content. Additionally, it will stress the urgency of accountability mechanisms for tech companies and institutions to ensure ethical AI development and application.
[…][…]
Bridging the Compute Divide: Evaluating NAIRR’s Role in Democratizing AI Research and Development
This project aims to evaluate the National Artificial Intelligence Research Resource (NAIRR) and its potential to democratize AI research and development (R&D) by providing underfunded American researchers and educators with access to essential computing resources. The NAIRR addresses the “compute divide,” which currently favors large technology firms and elite universities due to their substantial financial means. With a proposed budget of $2.25 billion over six years for the NAIRR, this evaluation will assess whether NAIRR can offer the necessary resources to enhance AI R&D, comparing its effectiveness to that of wealthier institutions. The ultimate goal is to determine NAIRR’s capability to bridge the compute gap and foster advanced AI model development.
[…][…]
Analysis of the Applicability of Effectiveness Frameworks in Development and Social Action Initiatives
The project idea arises from the awareness that many development aid initiatives are not effectively functioning. However, there is existing knowledge on how to conduct effective interventions and make optimal institutional decisions. The project aims to answer the question: Can the academic frameworks developed by organizations aligned with Effective Altruism be applied to projects by smaller actors? What is the best way to disseminate this knowledge? This idea is envisioned as a bridge, a consultancy work connecting academic frameworks with local initiatives, with the goal of finding the best way to exchange information between them.
[…][…]
Artificial Intelligence and Biometric Data Protection in Latin America: Lessons from the European Union
This project focuses on analyzing the risks associated with the use of Artificial Intelligence (AI) in Latin America, especially in relation to the protection of biometric data. Regional challenges will be examined, taking as a reference the legal framework and technologies used in the European Union to safeguard this data. Tools such as advanced encryption systems and Biometric Privacy Enhancement Techniques (B-PET) will be identified. Based on this analysis, a guide of recommendations adapted to the needs of Latin America will be developed, with the aim of strengthening the protection of biometric data and promoting best practices in AI and information security.
[…][…]
Visit One-Pager Paths of Fire: Mapping Historical Global Supervolcanic Eruptions
A major volcanic eruption poses a threat to the environment, food security, public health, and the socio-economic stability of a region. Although the frequency of a supereruption is difficult to calculate due to information bias and the number of unidentified supervolcanoes, it has been concluded that the probability of such an event occurring is approximately 1 in every 6 centuries. These findings indicate that humanity could face the impacts of a supereruption in the near future, especially given the lack of global efforts to mitigate its effects. Exploring supervolcanoes is useful for understanding their functioning and evolution over geological time. The proposed outreach article aims to address this need by mapping the major global supervolcanic eruptions that have marked Earth’s geological history, highlighting their short- and long-term impacts and consequences. The purpose is to raise awareness among decision-makers and the community about the importance of preparing for extreme scenarios and strengthening regional resilience to large-scale natural events.
[…][…]
Proposal For Prevention And Detection Of Pathogens With Pandemic Potential In Ecuador
This report aims to explore the challenges and opportunities in preventing and detecting pandemic-potential pathogens in Ecuador. It starts with a detailed analysis of the country’s current epidemiological situation, reviewing major biological threats and their potential public health impacts. It then examines existing surveillance systems and outlines the procedures and manuals for efficient outbreak response. Finally, it explores improvement opportunities in pathogen prevention and detection, proposing specific recommendations to strengthen health and surveillance systems, enhance outbreak response capabilities, and promote collaboration among various stakeholders in national and international risk management.
Gabriela was accepted in the Genomics and Epidemiological Surveillance of Bacterial Pathogens course from the Universidad de Costa Rica, after being recommended by Carreras con Impacto.
[…][…]
Legal and Ethical Implications of Simulating Human Emotions by AI
The project aims to gather information on the simulation of human emotions by artificial intelligence, resulting in a brief report on the legal and ethical implications. It will propose best practices for AI developers to guide responsible use and mitigate human-created biases, protecting individuals’ rights and ensuring transparency and accountability in using this technology. The current issue is the general public’s lack of understanding of AI’s potential in emulating human emotions and the possibility of manipulative use. Specialists from the World Compliance Association highlight concerns such as loss of control over personal information and the ability to make informed decisions. The research will employ a qualitative approach, collecting literature and conducting semi-structured interviews with experts. The final report will compare and analyze expert responses and literature review findings, to be published on Effective Altruism blogs, guiding government decision-makers in creating or modifying public policies and raising awareness among future AI developers to ensure ethical regulation in AI applications.
[…][…]
Epidemiological Library Project
The project aims to address biological risks by recording and analyzing responses to significant epidemiological outbreaks in Mexico over the past 25 years. By organizing and understanding this data, the project seeks to identify patterns and enhance the ability to anticipate and respond to future public health emergencies. Mexico’s diverse epidemiological history and large population make it an ideal focus. The collected data will be accessible through an interactive webpage, benefiting biosafety professionals by providing valuable information for research, data analysis, and decision-making. Comparing Mexico’s responses with those of other countries will also help develop tailored strategies for different contexts.
[…][…]
Exploring the consequences of Russia's absence from the New START Treaty for strategic arms reduction.
The project will investigate the implications of Russia’s withdrawal from the new START treaty with the US, analyzing the geopolitical impact on international security and the reasons behind Russia’s non-compliance. It aims to propose better negotiation strategies by evaluating sections and articles of the treaty and exploring diplomatic alternatives, including other bilateral or multilateral agreements and confidence-building measures among nuclear powers.
[…][…]
Randomized trial protocol based on Difference-In-Difference methodology for measuring the impact of JPAL LAC.
This project will estimate the impact of opening a research office in a developing country, specifically analyzing the JPAL office in Chile, using Randomized Controlled Trials (RCTs). It will identify similar countries for comparison based on criteria like public policy evidence usage and PISA scores. The secondary goal will be to develop a decision-making protocol for the cost-effectiveness of research office locations, creating a “Weighted Factors Model” based on the findings from the initial study.
[…][…]
ProteinAid India: Sustainable Nutritional Strategies
The project will examinate the feasibility and effectiveness of integrating alternative protein and plant-based diets as strategies to combat malnutrition and undernutrition in India.evaluating sections and articles of the treaty and exploring diplomatic alternatives, including other bilateral or multilateral agreements and confidence-building measures among nuclear powers.
[…][…]
A Journey through the Virus: Pandemic Awareness Documentary on Social Networks
This project aims to raise awareness about pandemic prevention through an audiovisual communication strategy on platforms like TikTok and Instagram Reels. The series will cover historical and current pandemics, combining historical elements, expert interviews, and impactful visual content. The methodology includes thorough research, detailed scripts, and audiovisual production. The findings will focus on the effectiveness of social media interventions in influencing audience behavior. In conclusion, this project seeks to fill an informational gap and contribute to future pandemic prevention while advancing the student creator’s career by integrating skills in communication, audiovisual production, and technology.
[…][…]
Podcast "Guardianes de Derecho": Explorando los Riesgos Catastróficos Globales
This proposal suggests that disseminating information about the role of lawyers in managing catastrophic risks in Latin America through a podcast will significantly increase awareness and sensitize society to the importance of legal professionals in disaster prevention and mitigation.
[…][…]
A Journey through the Virus: Pandemic Awareness Documentary on Social Networks
This project will explore the transformative impact of effective social media communication on the perceptions, attitudes, and behaviors of the Latin American audience regarding priority causes. Focusing on platforms like LinkedIn and Instagram, the research will target Latin American university students aged 18 to 26, aiming to increase their awareness of global catastrophic risks and important social issues through engaging content. The main objective is to demonstrate how continuous exposure to relevant social media content can raise public awareness. Initial strategies will include educating the audience on content and posting times, with a gradual increase in reach through tactics to boost followers and visibility.
[…][…]
Competence in Scalability and Performance of AI Models Based on Benchmarks Between the United States and China
This project will evaluate China’s progress in developing large-scale AI models and compare them with similar developments in the U.S., especially in terms of performance on benchmarks. This evaluation includes studying advances in infrastructure in both countries, the creation of Chinese benchmarks as new performance standards, development in each country, and a comparison of specific models. Additionally, the potential positive and negative impacts in various fields and the potential risks in the international arena will be analyzed.
[…][…]
Mapping the Presence of Antibiotic-Resistant Bacteria Using GIS and Medical Reports
Antibiotic resistance in bacteria is a major public health problem worldwide. Several organizations around the world are dedicated to tracking antibiotic-resistant bacteria, but these agencies invest relatively little compared to the magnitude of the problem, especially in low-income countries. There is a need for information collection and visualization tools that enable informed decision-making in the development of public policies to address this issue. The proposed solution involves a georeferencing system that integrates data from relevant agencies into geographic information systems (GIS) to identify areas of high incidence, potential infection hotspots, and, based on healthcare coverage, areas where the population is most vulnerable.
[…][…]
RoboTrending: Analysis of AI Trends in Robotics with a Focus on Reinforcement Learning
This project aims to identify and analyze trends in artificial intelligence models applied to robotics, with a special emphasis on reinforcement learning (RL). This will be achieved by using databases such as Epoch, IEEE Xplore, Scopus, Google Scholar, arXiv, Web of Science, etc. Through bibliometric and content analysis, the study will examine publication and collaboration patterns, as well as predominant topics and methods. Additionally, the evolution and scalability of these models will be compared with those in the field of language using predictive analysis techniques. The goal is to provide an informed basis for applying regulations that promote innovation and ensure the ethical development of technology.
[…][…]
Competence in Scalability and Performance of AI Models Based on Benchmarks Between the United States and China
This aims to conduct a comprehensive and comparative analysis to inform and raise public awareness about the risks associated with Artificial Intelligence (AI) and the necessity of specific regulations. Utilizing a matrix of factors, the project will compare AI regulations across Latin American countries and the European Union, which has recently passed the AI Act. This matrix will explore key areas such as existing AI laws, investment levels in AI research, technological advancement, data protection policies, and leading national AI-using companies. The project seeks to highlight the importance of fairness and justice in AI algorithm design, data privacy, security, and the authenticity of AI-generated content. Additionally, it will stress the urgency of accountability mechanisms for tech companies and institutions to ensure ethical AI development and application.
[…][…]
Regulating Change: The Urgency of Standards for the Responsible Advancement of AI
This project investigates the growth trends in Artificial Intelligence (AI) models, focusing on the processing capacity and energy consumption of data centers worldwide. The project will analyze the factors influencing energy consumption, evaluate various data centers, and propose optimizations to improve energy efficiency without compromising performance. It will include a literature review, data analysis, predictions, risk assessments, and ultimately, the drafting of a report detailing the findings. This report will serve as an information source for responsible and sustainable management of AI in the future.
[…][…]
AI Capabilities in Graph Problem Solving
This project, wilI intend to propose graph problem-solving tasks for different models, taking the BIG-benchmark format as a starting point and using models such as BLOOM, GPT-3, GPT-4, T5, FLAN-T5, and BERT.
[…][…]
RoboTrending: Analysis of AI Trends in Robotics with a Focus on Reinforcement Learning
This project aims to identify and analyze trends in artificial intelligence models applied to robotics, with a special emphasis on reinforcement learning (RL). This will be achieved by using databases such as Epoch, IEEE Xplore, Scopus, Google Scholar, arXiv, Web of Science, etc. Through bibliometric and content analysis, the study will examine publication and collaboration patterns, as well as predominant topics and methods. Additionally, the evolution and scalability of these models will be compared with those in the field of language using predictive analysis techniques. The goal is to provide an informed basis for applying regulations that promote innovation and ensure the ethical development of technology.
[…][…]
Competence in Scalability and Performance of AI Models Based on Benchmarks Between the United States and China
This project will evaluate China’s progress in developing large-scale AI models and compare them with similar developments in the U.S., especially in terms of performance on benchmarks. This evaluation includes studying advances in infrastructure in both countries, the creation of Chinese benchmarks as new performance standards, development in each country, and a comparison of specific models. Additionally, the potential positive and negative impacts in various fields and the potential risks in the international arena will be analyzed.
[…][…]
Artificial Intelligence and Biometric Data Protection in Latin America: Lessons from the European Union
This project focuses on analyzing the risks associated with the use of Artificial Intelligence (AI) in Latin America, especially in relation to the protection of biometric data. Regional challenges will be examined, taking as a reference the legal framework and technologies used in the European Union to safeguard this data. Tools such as advanced encryption systems and Biometric Privacy Enhancement Techniques (B-PET) will be identified. Based on this analysis, a guide of recommendations adapted to the needs of Latin America will be developed, with the aim of strengthening the protection of biometric data and promoting best practices in AI and information security.
[…][…]
Bridging the Compute Divide: Evaluating NAIRR’s Role in Democratizing AI Research and Development
This project aims to evaluate the National Artificial Intelligence Research Resource (NAIRR) and its potential to democratize AI research and development (R&D) by providing underfunded American researchers and educators with access to essential computing resources. The NAIRR addresses the “compute divide,” which currently favors large technology firms and elite universities due to their substantial financial means. With a proposed budget of $2.25 billion over six years for the NAIRR, this evaluation will assess whether NAIRR can offer the necessary resources to enhance AI R&D, comparing its effectiveness to that of wealthier institutions. The ultimate goal is to determine NAIRR’s capability to bridge the compute gap and foster advanced AI model development.
[…][…]
Open-source Replication of Decomposition of Language Models with Dictionary Learning
This project will replicate Anthropic’s paper on Dictionary Learning for feature extraction from Language Models [Bri+23], and explore potential extensions if possible.
⭐ Fernando was accepted to participate on the ML4Good Bootcamp during his mentorship in Carreras con Impacto.
[…][…]
Mapping the Presence of Antibiotic-Resistant Bacteria Using GIS and Medical Reports
Antibiotic resistance in bacteria is a major public health problem worldwide. Several organizations around the world are dedicated to tracking antibiotic-resistant bacteria, but these agencies invest relatively little compared to the magnitude of the problem, especially in low-income countries. There is a need for information collection and visualization tools that enable informed decision-making in the development of public policies to address this issue. The proposed solution involves a georeferencing system that integrates data from relevant agencies into geographic information systems (GIS) to identify areas of high incidence, potential infection hotspots, and, based on healthcare coverage, areas where the population is most vulnerable.
[…][…]
Epidemiological Library Project
The project aims to address biological risks by recording and analyzing responses to significant epidemiological outbreaks in Mexico over the past 25 years. By organizing and understanding this data, the project seeks to identify patterns and enhance the ability to anticipate and respond to future public health emergencies. Mexico’s diverse epidemiological history and large population make it an ideal focus. The collected data will be accessible through an interactive webpage, benefiting biosafety professionals by providing valuable information for research, data analysis, and decision-making. Comparing Mexico’s responses with those of other countries will also help develop tailored strategies for different contexts.
[…][…]
Proposal For Prevention And Detection Of Pathogens With Pandemic Potential In Ecuador
This report aims to explore the challenges and opportunities in preventing and detecting pandemic-potential pathogens in Ecuador. It starts with a detailed analysis of the country’s current epidemiological situation, reviewing major biological threats and their potential public health impacts. It then examines existing surveillance systems and outlines the procedures and manuals for efficient outbreak response. Finally, it explores improvement opportunities in pathogen prevention and detection, proposing specific recommendations to strengthen health and surveillance systems, enhance outbreak response capabilities, and promote collaboration among various stakeholders in national and international risk management.
Gabriela was accepted in the Genomics and Epidemiological Surveillance of Bacterial Pathogens course from the Universidad de Costa Rica, after being recommended by Carreras con Impacto.
[…][…]
Legal and Ethical Implications of Simulating Human Emotions by AI
The project aims to gather information on the simulation of human emotions by artificial intelligence, resulting in a brief report on the legal and ethical implications. It will propose best practices for AI developers to guide responsible use and mitigate human-created biases, protecting individuals’ rights and ensuring transparency and accountability in using this technology. The current issue is the general public’s lack of understanding of AI’s potential in emulating human emotions and the possibility of manipulative use. Specialists from the World Compliance Association highlight concerns such as loss of control over personal information and the ability to make informed decisions. The research will employ a qualitative approach, collecting literature and conducting semi-structured interviews with experts. The final report will compare and analyze expert responses and literature review findings, to be published on Effective Altruism blogs, guiding government decision-makers in creating or modifying public policies and raising awareness among future AI developers to ensure ethical regulation in AI applications.
[…][…]
Visit One-Pager Paths of Fire: Mapping Historical Global Supervolcanic Eruptions
A major volcanic eruption poses a threat to the environment, food security, public health, and the socio-economic stability of a region. Although the frequency of a supereruption is difficult to calculate due to information bias and the number of unidentified supervolcanoes, it has been concluded that the probability of such an event occurring is approximately 1 in every 6 centuries. These findings indicate that humanity could face the impacts of a supereruption in the near future, especially given the lack of global efforts to mitigate its effects. Exploring supervolcanoes is useful for understanding their functioning and evolution over geological time. The proposed outreach article aims to address this need by mapping the major global supervolcanic eruptions that have marked Earth’s geological history, highlighting their short- and long-term impacts and consequences. The purpose is to raise awareness among decision-makers and the community about the importance of preparing for extreme scenarios and strengthening regional resilience to large-scale natural events.
[…][…]
Analysis of the Applicability of Effectiveness Frameworks in Development and Social Action Initiatives
The project idea arises from the awareness that many development aid initiatives are not effectively functioning. However, there is existing knowledge on how to conduct effective interventions and make optimal institutional decisions. The project aims to answer the question: Can the academic frameworks developed by organizations aligned with Effective Altruism be applied to projects by smaller actors? What is the best way to disseminate this knowledge? This idea is envisioned as a bridge, a consultancy work connecting academic frameworks with local initiatives, with the goal of finding the best way to exchange information between them.
[…][…]
ProteinAid India: Sustainable Nutritional Strategies
The project will examinate the feasibility and effectiveness of integrating alternative protein and plant-based diets as strategies to combat malnutrition and undernutrition in India.evaluating sections and articles of the treaty and exploring diplomatic alternatives, including other bilateral or multilateral agreements and confidence-building measures among nuclear powers.
[…][…]
Randomized trial protocol based on Difference-In-Difference methodology for measuring the impact of JPAL LAC.
This project will estimate the impact of opening a research office in a developing country, specifically analyzing the JPAL office in Chile, using Randomized Controlled Trials (RCTs). It will identify similar countries for comparison based on criteria like public policy evidence usage and PISA scores. The secondary goal will be to develop a decision-making protocol for the cost-effectiveness of research office locations, creating a “Weighted Factors Model” based on the findings from the initial study.
[…][…]
Exploring the consequences of Russia's absence from the New START Treaty for strategic arms reduction.
The project will investigate the implications of Russia’s withdrawal from the new START treaty with the US, analyzing the geopolitical impact on international security and the reasons behind Russia’s non-compliance. It aims to propose better negotiation strategies by evaluating sections and articles of the treaty and exploring diplomatic alternatives, including other bilateral or multilateral agreements and confidence-building measures among nuclear powers.
[…][…]
A Journey through the Virus: Pandemic Awareness Documentary on Social Networks
This project will explore the transformative impact of effective social media communication on the perceptions, attitudes, and behaviors of the Latin American audience regarding priority causes. Focusing on platforms like LinkedIn and Instagram, the research will target Latin American university students aged 18 to 26, aiming to increase their awareness of global catastrophic risks and important social issues through engaging content. The main objective is to demonstrate how continuous exposure to relevant social media content can raise public awareness. Initial strategies will include educating the audience on content and posting times, with a gradual increase in reach through tactics to boost followers and visibility.
[…][…]
Podcast "Guardianes de Derecho": Explorando los Riesgos Catastróficos Globales
This proposal suggests that disseminating information about the role of lawyers in managing catastrophic risks in Latin America through a podcast will significantly increase awareness and sensitize society to the importance of legal professionals in disaster prevention and mitigation.
[…][…]
A Journey through the Virus: Pandemic Awareness Documentary on Social Networks
This project aims to raise awareness about pandemic prevention through an audiovisual communication strategy on platforms like TikTok and Instagram Reels. The series will cover historical and current pandemics, combining historical elements, expert interviews, and impactful visual content. The methodology includes thorough research, detailed scripts, and audiovisual production. The findings will focus on the effectiveness of social media interventions in influencing audience behavior. In conclusion, this project seeks to fill an informational gap and contribute to future pandemic prevention while advancing the student creator’s career by integrating skills in communication, audiovisual production, and technology.
[…][…]