The UN will regulate artificial intelligence
Robo advisor with AI arrives in Mexico. The dangers of malvertising and vishing in Latam. AI could help us to talk with animals.
The latest news and trends in the Latin American AI startup scene.
Share your stories to ai@contxto.com.
_
How the U.N. Plans to Shape the Future of AI
The United Nations (UN) is considering the establishment of a new agency to govern artificial intelligence (AI), according to the UN Secretary-General's envoy on technology, Amandeep Gill. The agency would address the risks and challenges posed by AI and promote international cooperation on AI issues.
The UN's High-Level Advisory Body on Artificial Intelligence is expected to produce reports on options for international AI governance, with the potential for recommending the establishment of a new agency. Different models for AI governance, including an International Atomic Energy Agency-like approach, an IPCC-like organization, a public-private partnership, or an international research effort, are being considered.
However, challenges in gaining support from countries like the US, China, and Russia remain. Some experts suggest starting with a less intrusive governance model, such as an International AI Organization partnering with national regulators. Despite the challenges, Gill believes there is a window of opportunity and that efforts to manage AI internationally should be pursued. (Time Magazine).
When it comes to creative thinking, it’s clear that AI systems mean business
Large language models (LLMs) like GPT-4 are considered "stochastic parrots" that generate sentences without understanding their meaning. However, they have shown surprising and unpredictable behaviors, leading to the question of whether they can be creative. In an experiment, GPT-4 outperformed humans in generating product ideas and received higher purchase intent ratings.
The study also highlighted the economic benefits of using LLMs, as they were found to be 40 times more productive and cost-effective than humans. This suggests that corporations may view LLMs as a valuable tool, similar to management consulting firms like McKinsey.
The essay raises the question of how to prevent AI from becoming another version of McKinsey, used to justify unpopular decisions. (The Guardian)
Quick News
Mexican startup Allie is one of the 20 most promising in the world
Mexican startup Allie Systems has been recognized as one of the top startups worldwide and has been selected to participate in TechCrunch Disrupt 2023.
The event provides an opportunity for the company to showcase its AI-based B2B workflow automation innovations and connect with investors.
Allie Systems, founded by former members of Rappi and Google, focuses on improving processes in the heavy machinery and manufacturing sector using IoT, AI, and machine learning. (Contxto)
Migtra and its Artificial Intelligence solutions arrive in Colombia
Chilean startup Migtra has expanded into the Colombian market with its AI-based transportation solutions.
The company's technologies aim to improve efficiency and prevent accidents in industrial transportation systems.
Migtra has already implemented its solutions in sectors such as mining and logistics and is in discussions with key players in Colombia, including Ecopetrol. (Contxto)
“Pregúntale a Fintual”: The Virtual Financial Revolution Arrives in Mexico
Fintech company Fintual has introduced a virtual financial advisor called "Pregúntale a Fintual" to improve financial education and access to financial products in Mexico.
The AI tool provides 24/7 answers to questions about money, investments, and financial products, aiming to empower users to make informed decisions.
This tool could significantly impact how Mexicans manage their finances and contribute to a more economically healthy society. (Contxto)
Uruguay to Digitally Transform Fiscal Management with IDB Support
Uruguay is set to undergo a digital transformation of its fiscal management with the support of a $20 million loan from the Inter-American Development Bank (IDB).
The program aims to improve fiscal management by streamlining spending, public procurement, and tax collection. It will enhance the Integrated Financial Information System and the Government Resource Planning Platform to better manage the national budget and administrative processes.
The loan will also fund a new technological platform for public procurement, improve data management tasks using AI, and enhance government revenue management. This digitalization will benefit the government, population, and businesses by lowering costs, increasing transparency, and providing better digital services. (Inter-American Development Bank)
Accelerating Mexico’s Digitalization, AI Regulation
Mexico's healthcare industry is harnessing AI-powered tools to improve resource management and data-driven diagnostics, potentially saving millions of dollars.
Integration of emerging technologies and innovative operational strategies is crucial for organizations to achieve effective digital transformations in Mexico. (Mexico Business News)
The sixth Latin America, United States, and Spain in the Global Economy forum took place in New York, focusing on pressing challenges such as artificial intelligence, food security, climate change, and economic recovery in Latin America.
Discussions also highlighted the need for green transformation and gender equality. The forum concluded with discussions on the Inflation Reduction Act and its impact on the U.S. economy's fight against climate change. (El País)
Appgate Identifies Malvertising Surge in Latin America
Malvertising, the use of malicious code in seemingly legitimate digital advertisements, is becoming a major threat to businesses in Latin America.
These ads redirect users to phishing websites and can deliver various malware variants. The realistic design of these ads makes them difficult to identify as malicious.
Appgate has developed a tool to detect and report phishing incidents from digital ads on search engines like Google, Yahoo, and Bing. This approach can help mitigate the proliferation of illegitimate advertisements in Latin America and protect against cyberattacks (Mexico Business News)
Marina Núñez Bespalova to Keynote CC Global Summit 2023
The 2023 CC Global Summit, to be held in Mexico City, has announced Marina Núñez Bespalova, Mexico's Undersecretary of Cultural Development, as the opening keynote speaker.
The Summit's theme of AI and the commons will challenge attendees to explore the intersection of artificial intelligence with open knowledge and culture.
The Summit aims to foster better sharing by cultivating a strategy that is contextual, inclusive, just, equitable, reciprocal, and sustainable. (Creative Commons)
Artificial Intelligence Enhances Vishing Attacks
AI is being used by cybercriminals to carry out more sophisticated social engineering attacks, such as vishing (voice phishing).
AI-powered tools can convincingly replicate a user's voice, making it easier for scammers to impersonate legitimate organizations and deceive people into providing personal information.
This trend is concerning, especially for countries like Mexico where phishing scams are already on the rise. To mitigate the risks, users are advised to remain vigilant for distorted voice quality and be skeptical of phone call promotions that seem too good to be true. (Mexico Business News)
Wayra to invest USD$1.4 million in Latin American startups
Telefónica Movistar's open innovation and corporate investment fund, Wayra Hispam, has added Colombian startup Celes to its portfolio. Celes uses artificial intelligence and data analysis to improve the cash flow of businesses.
Wayra Hispam has invested around $20 million in startups over the past 12 years and currently has over 100 active startups in its portfolio.
The fund plans to invest about $1.4 million in Hispanic American startups in 2023. The success of Chilean startup Webdox, which focuses on contract management and signing, was also highlighted. (Contxto)
AI Education
TinyML and Efficient Deep Learning Computing
This course focuses on efficient AI computing techniques for resource-constrained devices. It covers topics such as model compression, pruning, quantization, and neural architecture search.
The course also explores application-specific acceleration techniques for large language models, diffusion models, video recognition, and point cloud. Quantum machine learning is also discussed.
A Comprehensive Guide for Building RAG-based LLM Applications
In this guide, we're going to build a RAG-based LLM application where we will incorporate external data sources to augment our LLM’s capabilities. Specifically, we will be building an assistant that can answer questions about Ray — a Python framework for productionizing and scaling ML workloads. The goal here is to make it easier for developers to adopt Ray, but also, as we'll see in this guide, to help improve our Ray documentation itself and provide a foundation for other LLM applications. We’ll also share challenges we faced along the way and how we overcame them.
Validating Large Language Model Outputs
This article discusses the need for output validation in large language models (LLMs) and introduces Guardrails AI, an open-source package that helps implement output validation. LLMs produce probabilistic outputs, which require validation to ensure consistency and reliability.
The article provides examples of when output validation is needed, such as ensuring structure compliance, safe responses, semantic similarity, valid choices, and quality assurance. It then explains how to implement output validation using Guardrails AI, including creating an output schema, initializing a Guard object, and wrapping an LLM call with the Guard object.
Food for Thought
AI’s $200B Question
The rapid growth of the Generative AI wave, fueled by Nvidia's success, has led to a surge in investments and valuations. However, the article raises questions about the purpose of this massive investment in GPUs and who the end customers are. It highlights the need for generating significant revenue to justify the capital investment and discusses the role of big tech companies in driving data center build-outs.
The article also emphasizes the importance of creating real end-customer value and shifting focus towards leveraging AI technology to improve people's lives. It suggests that there is an opportunity for startups to fill the gap in revenue generation and calls for a shift in thinking from infrastructure to customer-centric value. (Sequoia Capital)
Artificial Intelligence Could Finally Let Us Talk with Animals
New technology powered by AI is revolutionizing the study of animal communication. AI is being used to decode and understand the vocalizations of animals such as crows and sperm whales. Researchers are collecting vast amounts of data and using machine learning models to analyze them.
AI has the potential to aid conservation efforts, improve animal welfare, and reshape our understanding of animals. AI models can identify individual animals based on their vocalizations, separate overlapping sounds, and even predict what words come next in a sequence. This technology opens up new possibilities for understanding and communicating with animals. (Scientific American)
The Growing Importance of AI Governance
AI governance is crucial for the safe and ethical implementation of AI technology. Scientists and researchers are joining forces with policymakers to emphasize the importance of AI governance as both an ethical code and a regulatory framework. The goal of AI governance is to ensure fair and equitable access to the benefits of AI while promoting transparency, safety, accountability, and the elimination of bias.
Generative AI, which creates new content based on existing data, poses risks such as job displacement, fake content creation, and the potential for AI systems to become sentient. To address these concerns, guidelines and regulations are being developed by technology firms and policymakers. The U.S. Office of Science and Technology Policy has proposed an AI Bill of Rights, while the European Union has proposed the Artificial Intelligence Act. Businesses can adopt a four-pronged strategy for AI governance, including reviewing and documenting AI usage and creating an AI monitoring system. (Dataversity)
_
Jobs
EY - GDS Consulting - GDS Data & Analytics Data Engineer - Senior 1- EY (Mexico City, Hybrid)
Data Scientist - Capgemini Engineering (Mexico, Hybrid)
Sr Developer - PepsiCo (Mexico City, Hybrid)