ibl.ai

OpenAI: Multi-Agent Portfolio Collaboration with OpenAI Agents SDK

Summary of https://cookbook.openai.com/examples/agents_sdk/multi-agent-portfolio-collaboration/multi_agent_portfolio_collaboration
Introduces a multi-agent system built using the OpenAI Agents SDK for complex investment research. It outlines an "agent as a tool" pattern where a central Portfolio Manager agent orchestrates specialized agents (Fundamental, Macro, Quantitative) and various tools to analyze market data and generate investment reports.
The text highlights the modularity, parallelism, and transparency offered by this architecture for building robust and scalable agent workflows. It details the different tool types supported by the SDK and provides an example output of the system in action, emphasizing the importance of structured prompts and tracing for building effective agent systems.
Complex tasks can be broken down and delegated to multiple specialist agents for deeper, higher-quality results. Instead of using a single agent for everything, multi-agent collaboration allows different autonomous agents to handle specific subtasks or expertise areas. In the investment research example, specialists like Macro, Fundamental, and Quantitative agents contribute their expertise, leading to a more nuanced and robust answer synthesized by a Portfolio Manager agent.
The "Agent as a Tool" pattern is a powerful approach for transparent and scalable multi-agent systems. This model involves a central agent (like the Portfolio Manager) calling other agents as tools for specific subtasks, maintaining a single thread of control and simplifying coordination. This approach is used in the provided example and allows for parallel execution of sub-tasks, making the overall reasoning transparent and auditable.
The OpenAI Agents SDK supports a variety of tool types, offering flexibility in extending agent capabilities.Agents can leverage built-in managed tools like Code Interpreter and WebSearch, connect to external services via MCP servers (like for Yahoo Finance data), and use custom Python functions (like for FRED economic data or file operations) defined with the function_tool decorator. This broad tool support allows agents to perform advanced actions and access domain-specific data.
Structured prompts and careful orchestration are crucial for building robust and consistent multi-agent workflows. The Head Portfolio Manager agent's system prompt encodes the firm's philosophy, tool usage rules, and a step-by-step workflow, ensuring consistency and auditability across runs. Modularity, parallel execution (enabled by features like parallel_tool_calls=True), and clear tool definitions are highlighted as best practices enabled by the SDK.
The system design emphasizes modularity, extensibility, and observability. By wrapping specialist agents as callable tools and structuring the workflow with a central coordinator, it's easier to update, test, or add new agents or tools. OpenAI Traces provide detailed visibility into every agent and tool call, making the workflow fully transparent and easier to debug.

Mary Meeker: Trends - Artificial Intelligence 2025

Summary of https://www.bondcap.com/report/pdf/Trends_Artificial_Intelligence.pdf
Extensively examines the rapid evolution of Artificial Intelligence, highlighting its unprecedented growth in user adoption, usage, and capital expenditure.
It details the competitive landscape, noting the rise of open-source models and the significant presence of China alongside the USA in AI development.
The text also explores AI's increasing integration into the physical world, its impact on workforces, and the ongoing investment in infrastructure like data centers and chips necessary to support this technological advancement.
The pace of change catalyzed by AI is unprecedented, ramping materially faster than the Internet's early growth. This is demonstrated by record-breaking user and usage growth for AI products like ChatGPT, which reached 800 million weekly active users in just 17 months, and significantly faster user adoption compared to previous technologies. Capital expenditure (CapEx) by major technology companies is also growing rapidly, increasingly directed towards building AI infrastructure like data centers and specialized hardware.
A key economic dynamic in AI is the tension between high and rising model training costs and rapidly falling inference costs per token. While training a frontier AI model can cost hundreds of millions or potentially billions of dollars, the cost to run these models (inference) has plummeted, with energy required per token falling drastically due to hardware and algorithmic advancements. This cost reduction is increasing accessibility and driving rising developer usage and new product creation, but also raises questions about the monetization and profitability of general-purpose LLMs.
The AI landscape is marked by rising competition among tech incumbents, emerging attackers, and global powers. Key threats to monetization include this intense competition, the growing capabilities and accessibility of open-source models which are closing the performance gap with closed models, and the rapid advancement and relevance of China's AI capabilities, which are catching up to USA models, increasingly powered by local semiconductors, and dominating domestic usage.
AI adoption and evolution are happening across diverse sectors and applications at a rapid pace. Beyond digital applications, AI is increasingly integrating into the physical world, enabling autonomous systems in areas like transportation, defense, agriculture, and robotics. It is also fundamentally transforming work, driving productivity improvements for employees and leading to significant growth in AI-related job postings and the adoption of AI tools by firms.
AI is poised to fundamentally reshape the internet experience for the next wave of global users, who may come online through AI-native interfaces (like conversational agents) powered by expanding satellite connectivity, potentially bypassing traditional app ecosystems. This technological shift is intertwined with increasing geopolitical competition, particularly between the United States and China, where leadership in AI is viewed as a critical component of national resilience and geopolitical influence, creating an AI "space race" with significant international implications.

Center for AI Policy: AI Agents – Governing Autonomy in the Digital Age

Summary of https://cdn.prod.website-files.com/65af2088cac9fb1fb621091f/682f96d6b3bd5a3e1852a16a_AI_Agents_Report.pdf
Presents an overview of AI agents, defined as autonomous systems capable of complex tasks without constant human supervision, highlighting their rapid progression from research to real-world application.
It identifies three major risks: catastrophic misuse through malicious applications, gradual human disempowerment as decision-making shifts to algorithms, and significant workforce displacement due to automation of cognitive tasks.
The report proposes four policy recommendations for Congress, including an Autonomy Passport for registration and oversight, mandatory continuous monitoring and recall authority, requiring human oversight for high-consequence decisions, and implementing workforce impact research to address potential job losses. These measures aim to mitigate the risks while allowing the beneficial aspects of AI agent development to continue.
AI agents represent a significant shift in AI capabilities, moving from research to widespread deployment. Unlike chatbots, these systems are autonomous and goal-directed, capable of taking a broad objective, planning their own steps, using external tools, and iterating without continuous human prompting. They can operate across multiple digital environments and automate decisions, not just steps. Agent autonomy exists on a spectrum, categorized into five levels ranging from shift-length assistants to frontier super-capable systems.
The widespread adoption of autonomous AI agents presents three primary risks: catastrophic misuse, where agents could enable dangerous attacks or cyber-intrusions; gradual human disempowerment, as decision-making power shifts to opaque algorithms across economic, cultural, and governmental systems; and workforce displacement, with projections indicating that tasks equivalent to roughly 300 million full-time global positions could be automated, affecting mid-skill and cognitive roles more rapidly than previous automation waves.
To mitigate these risks, the report proposes four key policy recommendations for Congress. These include creating a federal Autonomy Passport system for registering high-capability agents before deployment, mandating continuous oversight and recall authority (including containment and provenance tracking) to quickly suspend problematic deployments, requiring human oversight by qualified professionals for high-consequence decisions in domains like healthcare, finance, and critical infrastructure, and directing federal agencies to monitor workforce impacts annually.
The proposed policy measures are designed to be proportional to the level of agent autonomy and the domain of deployment, focusing rigorous oversight on where autonomy creates the highest risk while allowing lower-risk innovation to proceed. For instance, the Autonomy Passport requirement and continuous oversight mechanisms target agents classified at Level 2 or higher on the five-level autonomy scale.
Early deployments demonstrate significant productivity gains, and experts project agents could tackle projects equivalent to a full human work-month by 2029. However, the pace of AI agent development is accelerating faster than the governance frameworks designed to contain its risks, creating a critical mismatch and highlighting the need for proactive policy intervention before the next generation of agents is widely deployed.

North-West University: Exploring AI-Driven Conversations as Dynamic OER for Self-Directed Learners

Summary of https://conference.pixel-online.net/files/foe/ed0015/FP/8250-ESOC7276-FP-FOE15.pdf
This conceptual paper explores the potential of AI-driven conversations, such as those from ChatGPT, to function as dynamic Open Educational Resources (OER) that support self-directed learning (SDL).
Unlike traditional, static resources, AI-powered dialogues offer personalized, interactive, and adaptive experiences that align with learners' needs. The paper argues that these tools can nurture key SDL competencies while acknowledging ethical, pedagogical, and technical considerations.
Ultimately, the authors propose that thoughtfully designed AI-driven OER can empower learners and teachers and contribute to a more inclusive and responsive future for open education.
AI-driven conversations can act as dynamic OER to support SDL. AI-driven conversations, such as those facilitated by ChatGPT, have the potential to function as dynamic Open Educational Resources (OER). Unlike traditional static resources, these dialogues offer personalised, interactive, and adaptive experiences that align with learners' unique needs and goals. This dynamic capability contrasts with static OER.
AI supports core principles and competencies of Self-Directed Learning (SDL). AI-driven conversations and generative AI tools can nurture key SDL competencies such as goal setting, self-monitoring, and reflective practice. They support learner autonomy, responsibility, self-motivation, and empower students to take initiative, plan, and manage their learning processes. AI also enhances online collaboration, creativity, problem-solving, and communication skills, which align with SDL characteristics.
AI integration can enhance Open Educational Practices (OEP) and improve access and inclusivity.Integrating AI into OEP holds the potential to address long-standing challenges in open education, such as learner engagement, the wider reach and adaptability of resources, and inclusive access. AI supports the creation of diverse and inclusive learning resources, facilitating multilingual and culturally relevant content generation. This integration aligns with the values of access, equity, and transparency that underpin open education.
Significant challenges exist in integrating AI into open education. Key challenges include legal and ethical concerns related to copyright, data privacy, and potential biases in AI outputs. There are also technical limitationsdue to fragmented OER infrastructure and a critical need for teacher preparedness and AI literacy, as many educators lack the foundational knowledge and confidence to use AI technologies effectively.
Successful integration requires thoughtful planning, policy, and professional development. To effectively realise the potential of AI-driven OER for SDL within OEP, it requires thoughtful design, robust infrastructure, inclusive policies, and sustained professional development for teachers. Recommendations include developing ethical guidelines, investing in compatible OER infrastructure, promoting inclusive AI design, providing professional development focused on both AI literacy and SDL skills for teachers, and encouraging ongoing research.

Google: Agents Companion

Summary of https://www.kaggle.com/whitepaper-agent-companion
This technical document, the Agents Companion, explores the advancements in generative AI agents, highlighting their architecture composed of models, tools, and an orchestration layer, moving beyond traditional language models.
It emphasizes Agent Ops as crucial for operationalizing these agents, drawing parallels with DevOps and MLOps while addressing agent-specific needs like tool management.
The paper thoroughly examines agent evaluation methodologies, covering capability assessment, trajectory analysis, final response evaluation, and the importance of human-in-the-loop feedback alongside automated metrics. Furthermore, it discusses the benefits and challenges of multi-agent systems, outlining various design patterns and their application, particularly within automotive AI.
Finally, the Companion introduces Agentic RAG as an evolution in knowledge retrieval and presents Google Agentspace as a platform for developing and managing enterprise-level AI agents, even proposing the concept of "Contract adhering agents" for more robust task execution.
Agent Ops is Essential: Building successful agents requires more than just a proof-of-concept; it necessitates embracing Agent Ops principles, which integrate best practices from DevOps and MLOps, while also focusing on agent-specific elements such as tool management, orchestration, memory, and task decomposition.
Metrics Drive Improvement: To build, monitor, and compare agent revisions, it is critical to start with business-level Key Performance Indicators (KPIs) and then instrument agents to track granular metrics related to critical tasks, user interactions, and agent actions (traces). Human feedback is also invaluable for understanding where agents excel and need improvement.
Automated Evaluation is Key: Relying solely on manual testing is insufficient. Implementing automated evaluation frameworks is crucial to assess an agent's core capabilities, its trajectory (the steps taken to reach a solution, including tool use), and the quality of its final response. Techniques like exact match, in-order match, and precision/recall are useful for trajectory evaluation, while autoraters (LLMs acting as judges) can assess final response quality.
Human-in-the-Loop is Crucial: While automated metrics are powerful, human evaluation provides essential context, particularly for subjective aspects like creativity, common sense, and nuance. Human feedback should be used to calibrate and validate automated evaluation methods, ensuring alignment with desired outcomes and preventing the outsourcing of domain knowledge.
Multi-Agent Systems Offer Advantages: For complex tasks, consider leveraging multi-agent architectures. These systems can enhance accuracy through cross-checking, improve efficiency through parallel processing, better handle intricate problems by breaking them down, increase scalability by adding specialized agents, and improve fault tolerance. Understanding different design patterns like sequential, hierarchical, collaborative, and competitive is important for choosing the right architecture for a given application.

UC San Diego: Large Language Models Pass the Turing Test

Summary of https://arxiv.org/pdf/2503.23674
Researchers Cameron R. Jones and Benjamin K. Bergen investigated whether advanced large language models (LLMs) can pass the standard three-party Turing test. Their study involved human interrogators conversing with both a human and an AI, then judging which was human.
The findings indicate that GPT-4.5, when prompted to adopt a persona, was identified as human significantly more often than the actual human participant, marking the first empirical evidence of an AI passing this rigorous version of the test.
While other models like LLaMa-3.1 showed some human-like qualities, only the persona-prompted models consistently fooled human judges. The study also explored the strategies used by interrogators and the implications of these results for our understanding of AI capabilities and their societal impact.
This study provides the first empirical evidence that a Large Language Model (LLM), specifically GPT-4.5 when prompted to adopt a humanlike persona, can pass a standard three-party Turing test. In the study, GPT-4.5-PERSONA was judged as human 73% of the time, significantly more often than the actual human participants.
Prompting with a humanlike persona was crucial for success. While GPT-4.5-PERSONA and LLAMA-PERSONA were not reliably distinguished from humans, the same models without the persona prompt (GPT-4.5-NO-PERSONA and LLAMA-NO-PERSONA) performed significantly worse, often being identified as AI at rates significantly below chance in the undergraduate study.
The study compared the performance of several AI systems, including GPT-4.5, LLaMa-3.1-405B, GPT-4o, and ELIZA. The baseline models, GPT-4o-NO-PERSONA and ELIZA, had significantly lower win rates, indicating that interrogators could generally distinguish them from humans. This suggests the interrogators were not simply guessing randomly.
The research indicates that interrogators often relied on social, emotional, and linguistic cues rather than traditional measures of knowledge and reasoning when trying to distinguish between humans and AI. Interestingly, providing strange prompts or using "jailbreaks" was the most effective strategy for interrogators, while asking about the weather or human experiences was least effective.
The findings have significant social and economic implications, suggesting that contemporary LLMs could potentially substitute for humans in short conversations, raising concerns about deception, misinformation, and the potential undermining of real human interaction. The study also found that general knowledge about LLMs and frequent chatbot interaction did not consistently improve participants' ability to distinguish AI from humans.

Elon University: Being Human in 2035 – How Are We Changing in the Age of AI?

Summary of https://imaginingthedigitalfuture.org/wp-content/uploads/2025/03/Being-Human-in-2035-ITDF-report.pdf
This Elon University Imagining the Digital Future Center report compiles insights from a non-scientific canvassing of technology pioneers, builders, and analysts regarding the potential shifts in human capacities and behaviors by 2035 due to advanced AI. Experts anticipate blurred boundaries between reality and fiction, human and artificial intelligence, and human and synthetic creations, alongside concerns about eroding individual identity, autonomy, and critical thinking skills.
The report explores both optimistic visions of AI augmenting human potential and creativity and pessimistic scenarios involving increased dependence, social division, and the erosion of essential human qualities like empathy and moral judgment. Ultimately, it highlights the critical need for ethical development, regulation, and education to navigate the profound societal changes anticipated in the coming decade.
A significant majority of experts anticipate deep and meaningful or even fundamental and revolutionary change in people’s native operating systems and operations as humans broadly adapt to and use advanced AI by 2035.
Experts predict mostly negative changes in several core human traits and behaviors by 2035, including social and emotional intelligence, the capacity for deep thinking, trust in shared values, empathy, mental well-being, sense of agency, and sense of identity and purpose.
Conversely, pluralities of experts expect mostly positive changes in human curiosity and capacity to learn, decision-making and problem-solving abilities, and innovative thinking and creativity due to interactions with AI.
Many experts express concern about the potential for AI to be used in ways that de-augment humanity, serving the interests of tool builders and those in power, potentially leading to a global sociotechnical dystopia. However, they also see the potential for AI to augment human intelligence and bring about universal enlightenment if the direction of development changes.
The experts underscore the critical importance of how humans choose to integrate AI into their lives and societies. They emphasize the need for ethical considerations, human-centered design, the establishment of human values in AI development and policy, and the preservation of human agency to ensure AI serves humanity's flourishing rather than diminishing essential human capacities.

Bain & Company: Nvidia GTC 2025 – AI Matures into Enterprise Infrastructure

Summary of https://www.bain.com/globalassets/noindex/2025/bain_article_nvidia_gtc_2025_ai_matures_into_enterprise_infrastructure.pdf
Nvidia's GTC 2025 highlighted a significant shift in AI, moving from experimental phases to becoming core enterprise infrastructure. The event showcased how data remains crucial, but AI itself is now a data generator, leading to new insights and efficiencies.
Furthermore, smaller, specialized AI models are gaining prominence, offering cost advantages and improved control. While fully autonomous AI agents are still rare, structured semi-autonomous systems with human oversight are becoming standard.
Finally, the conference underscored the growing importance of digital twins, video analytics, and accessible off-the-shelf tools in democratizing enterprise AI adoption and fostering cross-functional collaboration through simulation.
AI has matured beyond pilot projects and is now being deployed at scale within the core operations of enterprises. Companies are re-architecting how they compete by moving AI from innovation teams into the business core.
Data remains both a critical challenge and a significant opportunity for AI success. Successful AI deployments rely on clean, connected, and accessible data. Furthermore, AI is now generating a new layer of data through insights and generative applications.
The trend is shifting towards smaller, specialized AI models that are more cost-effective and offer better control, latency, and privacy. Techniques like quantization, pruning, and RAG are facilitating this shift, although deploying and managing these custom models presents new operational complexities.
Agentic AI is gaining traction, but its successful implementation hinges on structure, transparency, and human oversight. While fully autonomous agents are rare, semiautonomous systems with built-in safeguards and orchestration platforms are becoming the near-term standard.
Digital twins and simulation have moved from innovation showcases to everyday enterprise tools, enabling faster rollout cycles, lower risk, and more informed decision-making. Simulation is also evolving into a collaboration platform for cross-functional teams.

Anthropic: Circuit Tracing – Revealing Computational Graphs in Language Models

Summary of https://transformer-circuits.pub/2025/attribution-graphs/methods.html
Introduces a novel methodology called "circuit tracing" to understand the inner workings of language models. The authors developed a technique using "replacement models" with interpretable components to map the computational steps of a language model as "attribution graphs." These graphs visually represent how different computational units, or "features," interact to process information and generate output for specific prompts.
The research details the construction, visualization, and validation of these graphs using an 18-layer model and offers a preview of their application to a more advanced model, Claude 3.5 Haiku. The study explores the interpretability and sufficiency of this method through various evaluations, including case studies on acronym generation and addition.
While acknowledging limitations like missing attention circuits and reconstruction errors, the authors propose circuit tracing as a significant step towards achieving mechanistic interpretability in large language models.
This paper introduces a methodology for revealing computational graphs in language models using Cross-Layer Transcoders (CLTs) to extract interpretable features and construct attribution graphs that depict how these features interact to produce model outputs for specific prompts. This approach aims to bridge the gap between raw neurons and high-level model behaviors by identifying meaningful building blocks and their interactions.
The methodology involves several key steps: training CLTs to reconstruct MLP outputs, building attribution graphs with nodes representing active features, tokens, errors, and logits, and edges representing linear effects between these nodes. A crucial aspect is achieving linearity in feature interactions by freezing attention patterns and normalization denominators. Attribution graphs allow for the study of how information flows from the input prompt through intermediate features to the final output token.
The paper demonstrates the application of this methodology through several case studies, including acronym generation, factual recall, and small number addition. These case studies illustrate how attribution graphs can reveal the specific features and pathways involved in different cognitive tasks performed by language models. For instance, in the addition case study, the method uncovers a hierarchy of heuristic features that collaboratively solve the task.
Despite the advancements, the methodology has several significant limitations. A key limitation is the missing explanation of how attention patterns are formed and how they mediate feature interactions (QK-circuits), as the analysis is conducted with fixed attention patterns. Other limitations include reconstruction errors (unexplained model computation), the role of inactive features and inhibitory circuits, the complexity of the resulting graphs, and the difficulty of understanding global circuits that generalize across many prompts.
The paper also explores the concept of global weights between features, which are prompt-independent and aim to capture general algorithms used by the replacement model. However, interpreting these global weights is challenging due to issues like interference (spurious connections) and the lack of accounting for attention-mediated interactions. While attribution graphs provide insights on specific prompts, future work aims to enhance the understanding of global mechanisms and address current limitations, potentially through advancements in dictionary learning and handling of attention mechanisms.

RAND: Uneven Adoption of AI Tools Among U.S. Teachers and Principals in the 2023-2024 School Year

Summary of https://www.rand.org/content/dam/rand/pubs/research_reports/RRA100/RRA134-25/RAND_RRA134-25.pdf
A RAND Corporation report, utilizing surveys from the 2023-2024 school year, investigates the adoption and use of artificial intelligence tools by K-12 public school teachers and principals. The research highlights that roughly one-quarter of teachers reported using AI for instructional planning or teaching, with higher usage among ELA and science teachers and those in lower-poverty schools.
Simultaneously, nearly 60 percent of principals indicated using AI in their jobs, primarily for administrative tasks like drafting communications. The study also found that guidance and support for AI use were less prevalent in higher-poverty schools for both educators, suggesting potential inequities in AI integration. Ultimately, the report underscores the emerging role of AI in education and recommends developing strategies and further research to ensure its effective and equitable implementation.
A significant portion of educators are using AI tools, but there's considerable variation. Approximately one-quarter of teachers reported using AI tools for instructional planning or teaching, with higher rates among ELA and science teachers, as well as secondary teachers. Notably, nearly 60 percent of principals reported using AI tools in their jobs. However, usage differed by subject taught and school characteristics, with teachers and principals in higher-poverty schools being less likely to report using AI tools.
Teachers primarily use AI for instructional planning, while principals focus on administrative tasks. Teachers most commonly reported using AI to generate lesson materials, assess students, and differentiate instruction. Principals primarily used AI to draft communications, support other school administrative tasks, and assist with teacher hiring, evaluation, or professional learning.
Disparities exist in AI adoption and support based on school poverty levels. Teachers and principals in lower-poverty schools were more likely to use AI and reported receiving more guidance on its use compared to their counterparts in higher-poverty schools. Furthermore, schools in higher-poverty areas were less likely to be developing AI usage policies. This suggests a widening gap in AI integration and the potential for unequal access to its benefits.
Educators have several concerns regarding AI use, including a lack of professional learning and data privacy. Principals identified a lack of professional development, concerns about data privacy, and uncertainty about how to use AI as major influences on their AI adoption. Teachers also expressed mixed perceptions about AI's helpfulness, noting the need to assess the quality of AI output and potential for errors.
The report highlights the need for intentional strategies and further research to effectively integrate AI in education. The authors recommend that districts and schools develop strategies to support AI use in ways that improve instruction and learning, focusing on AI's potential for differentiated instruction, practice opportunities, and student engagement. They also emphasize the importance of research to identify effective AI applications and address disparities in access and guidance, particularly for higher-poverty schools.