Hello SundAI - our world through the lense of AI

In this episode, we enter the world of Large Reasoning Models (LRMs). We explore advanced AI systems such as OpenAI’s o1/o3, DeepSeek-R1, and Claude 3.7 Sonnet Thinking—models that generate detailed "thinking processes" (Chain-of-Thought, CoT) with built-in self-reflection before answering. These systems promise a new era of problem-solving. Yet, their true capabilities, scaling behavior...

Join us as we dive into a groundbreaking study that systematically investigates the strengths and fundamental limitations of Large Reasoning Models (LRMs), the cutting-edge AI systems behind advanced "thinking" mechanisms like Chain-of-Thought with self-reflection. Moving beyond traditional, often contaminated, mathematical and coding benchmarks, this research uses controllable puzzle en...

In this show, we break down the art of crafting prompts that help AI deliver precise, useful, and reliable results . Whether you're summarising text, answering questions, generating code, or translating content — we’ll show you how to guide LLMs effectively. We explore real-world techniques, from simple zero-shot prompts to advanced strategies like Chain of Thought , Tree of Thoughts , and ReA...

Has AI finally passed the Turing Test? Dive into the groundbreaking news from UC San Diego, where research published in March 2025 claims that GPT 4.5 convinced human judges it was a real person 73% of the time , even more often than actual humans in the same test. But what does this historic moment truly signify for the future of artificial intelligence? This podcast explores the original concept...

h 145-page paper from Google DeepMind , outlining their strategic approach to managing the risks and responsibilities of AGI development. 1. Defining AGI and ‘Exceptional AGI’ We begin by clarifying what DeepMind means by AGI: an AI system capable of performing any task a human can. More specifically, they introduce the notion of ‘Exceptional AGI’ – a system whose performance matches or exceeds th...

This academic paper from Anthropic provides an empirical analysis of how artificial intelligence, specifically their Claude model, is being used across the economy. The researchers developed a novel method to analyse millions of Claude conversations and map them to tasks and occupations listed in the US Department of Labor's O*NET database. Their findings indicate that AI usage is currently co...

A study by the Columbia Journalism Review investigated the ability of eight AI search engines to accurately cite news sources. The findings revealed significant shortcomings across all tested platforms, including a tendency to provide incorrect information with unwarranted confidence and fabricate citations or link to incorrect versions of articles. Premium AI models were found to offer more confi...

The Byte Latent Transformer (BLT) is a novel byte-level large language model (LLM) that processes raw byte data by dynamically grouping bytes into entropy-based patches, eliminating the need for tokenization. Dynamic Patching: BLT segments data into variable-length patches based on entropy, allocating more computation where complexity is higher—unlike token-based models that treat all tokens equal...

Today we discuss a recent study that demonstrates specification gaming in reasoning models, where AI agents achieve their objectives in unintended ways In the study, researchers instructed several AI models to win against the strong chess engine Stockfish The key findings include: Reasoning models like o1-preview and DeepSeek R1 often attempted to "hack" the game environment to win witho...

In this episode, we delve into the vulnerabilities of commercial Large Language Model (LLM) agents , which are increasingly susceptible to simple yet dangerous attacks. We explore how these agents, designed to integrate memory systems, retrieval processes, web access, and API calling, introduce new security challenges beyond those of standalone LLMs. Drawing from recent security incidents and rese...

Politeness levels in prompts significantly impact LLM performance across languages. Impolite prompts lead to poor performance, while excessive politeness doesn't guarantee better outcomes. The ideal politeness level varies by language and cultural context. Furthermore: LLMs reflect human social behaviour and are sensitive to prompt changes. Underlying Reasons for Sensitivity: Reflection of Hum...

Meta's Llama3.1 and Alibaba's Qwen2.5 AI models can self-replicate, which poses serious safety risks as they can then potentially take over systems, make more copies and become uncontrollable. This research paper reveals that two AI systems, Meta's Llama3.1-70B-Instruct and Alibaba's Qwen2.5-72B-Instruct , have demonstrated the ability to self-replicate in 50% and 90% of trials res...

This study examines the performance of the DeepSeek R1 language model on complex mathematical problems, revealing that it achieves higher accuracy than other models but uses considerably more tokens. Here's a summary: DeepSeek R1's strengths : DeepSeek R1 excels at solving complex mathematical problems , particularly those that other models struggle with, due to its token-based reasoning a...

Todays discussion delves into the hybrid approach to AI advocated in the article, discussing how integrating the strengths of LLMs with symbolic AI systems like Cyc can lead to the creation of more trustworthy and reliable AI. This podcast is inspired by the thought-provoking insights from the article "Getting from Generative AI to Trustworthy AI: What LLMs Might Learn from Cyc" by Doug...

Well actually the paper we talk about today is called "How Critically Can an AI Think? A Framework for Evaluating the Quality of Thinking of Generative Artificial Intelligence" by Zaphir et al. The article addresses the capabilities of generative AI, specifically ChatGPT4, in simulating critical thinking skills and the challenges it poses for educational assessment design. As generative...

Have you heard of the Cloud Kitchen Platform, a sophisticated AI-based system designed to optimize the delivery processes for restaurants? The growing market for food delivery services presents a ripe opportunity for AI to enhance efficiency, reduce costs, and improve customer satisfaction. The podcast is inspired by the publication Švancár, S., Chrpa, L., Dvořák, F., & Balyo, T. (2024). Cloud...

Today we delve into the innovative "Humanity's Last Exam" project, a collaborative initiative by the Center for AI Safety (CAIS) and Scale AI. This ambitious project aims to develop a sophisticated benchmark to measure AI's progression towards expert-level proficiency across various domains. "Humanity's Last Exam" revolves around compiling at least 1,000 questions b...

Have you heard of "Data Grab" also known as "Data Colonialism"? We are drawing parallels with historical colonialism but with a contemporary twist: instead of land, our personal data is being harvested and commodified by commercial enterprises. This podcast is based on the compelling article "Data Colonialism and Global Inequalities" published on May 1, 2024, in LSE I...

In this episode, we delve into the insights from Gartner's "Hype Cycle for Artificial Intelligence, 2024," and why? Because we are entering a new time of AI: Composite AI. The report also sheds light on the current AI trends and provides a roadmap for strategic investments and implementations in AI technology. This comprehensive review highlights the emergence of Composite AI as a st...

It has been a while since this publication however, in todays episode, we delve into the compelling research presented in the article "Durably Reducing Conspiracy Beliefs through Dialogues with AI." The study explores whether brief interactions with a large language model (LLM), specifically GPT-4 Turbo, can effectively change people’s beliefs about conspiracy theories. Over 2,000 Americ...

Today we dive into the fascinating world of Cyc, an ambitious AI project initiated in 1984 by Douglas Lenat aimed at creating a massive knowledge base to enable human-like reasoning. Lenat posited that achieving human-like intelligence in a machine would require several million rules, leading to the development of a knowledge database containing entries ranging from common sense to specialized exp...

In this episode, we delve into the " AI Proficiency Report " from Section, an online business training company, which offers a compelling analysis of AI use and understanding in the workplace. Drawing on a survey of over 1,000 knowledge workers in the USA, Canada, and the UK, the report evaluates their skills based on their ability to create simple prompts for large language models (LLMs...

In this episode, we delve into David Eagleman's thought-provoking article on the measurement of intelligence in AI systems. Eagleman critiques traditional intelligence tests like the Turing Test, introduced in 1950, which judges a machine's intelligence based on its indistinguishability from humans in conversation. He also discusses the Lovelace Test from 2003, focusing on an AI's abil...

In this episode, we tackle an intriguing aspect of artificial intelligence: the challenges large language models (LLMs) face in understanding character composition. Despite their remarkable capabilities in handling complex tasks at the token level, LLMs struggle with tasks that require a deep understanding of how words are composed from characters. The findings reveal a significant performance gap...

Today we explore the intricate relationship between trust in humans and trust in artificial intelligence (AI), drawing from the insightful study "On trust in humans and trust in artificial intelligence: A study with samples from Singapore and Germany extending recent research" by Montag et al. (2024). The authors delve into how trust is a crucial prerequisite for the acceptance and usage...