Table of Contents

• This Week

• Business Use Cases

• Things to Pay Attention to

This Week

Bottom Line Up Front

• A data flywheel is crucial for businesses, and it needs to tie to a business metric to be effective, with the goal of building a moat that drives business growth.

• Nature abhors a vacuum. The market is experiencing a "tremendous sucking sound" for AI, driven by the rising tide of technology adoption, and companies should take advantage of this opportunity to grow at maximum velocity.

• The agent economy will require a shift in mindset, including a stochastic mindset that acknowledges the uncertainty and unpredictability of AI systems, a management mindset that understands the capabilities and limitations of agents, and a combination of these two mindsets to thrive in a world with more leverage but less certainty

Business Use Cases

Bottom Line Up Front

• Today, AWS launched findings from its global Gen AI Adoption Index research. This study surveyed 3,739 senior IT decision-makers in nine countries across industries such as financial services, information and communications technology, manufacturing, and retail. Survey regions include US, Brazil, Canada, France, Germany, India, Japan, South Korea, and the United Kingdom.

• 60% of organizations globally have appointed a dedicated AI executive, such as a Chief AI Officer, to drive adoption and manage implementation complexity.

• 92% of organizations are planning to recruit for roles requiring gen AI skills in 2025, adopting a dual strategy of developing internal workforce and external talent acquisition.

• A majority of organizations (58%) are taking a hybrid approach to gen AI adoption, developing custom applications using pre-existing models, while 55% will build applications on fine-tuned models using proprietary data.

Why is this important

• OpenAI is introducing "OpenAI for Countries", an initiative to help countries build their own AI infrastructure, similar to the Stargate project in the US.

• The initiative will provide partnerships to build in-country data center capacity, offer customized ChatGPT, and continue evolving security and safety controls for AI models.

• OpenAI aims to pursue 10 projects with individual countries or regions as the first phase of this initiative, promoting democratic AI and US-led AI leadership.Klarna CEO Reverses Course By Hiring More Humans, Not AI | Entrepreneur

Why is this important

• Klarna CEO Sebastian Siemiatkowski is reversing the company's AI-focused strategy, which had led to a hiring freeze, to hire more human staff for customer service.

• Siemiatkowski believes that having human support is critical for the company's brand and customer experience, despite AI being cheaper, as it provides a higher quality output.

• Klarna is now recruiting new customer service employees, targeting students, rural populations, and dedicated users, with a focus on providing quality human support for customers.

Bottom Line Up Front

• Google is using AI to combat online scams, detecting and blocking hundreds of millions of scammy results every day in Search.

• The company is utilizing AI-powered technologies, such as Gemini Nano, to enhance Safe Browsing in Chrome and protect users from phishing and other scams.

• Google is also launching AI-powered warnings for Chrome on Android to fight scams, spam, and unwanted notifications, and has introduced on-device AI-powered Scam Detection in Google Messages and Phone by Google to safeguard users from sophisticated call and messaging scams.

Things to Pay Attention to

Bottom Line Up Front - Which ChatGPT Model to use and when

• ChatGPT Enterprise offers access to various models, including GPT-4o, GPT-4.5, OpenAI o4-mini, OpenAI o4-mini-high, OpenAI o3, and OpenAI o1 pro mode.

• Each model has its own strengths, such as GPT-4o for everyday tasks, GPT-4.5 for creative tasks, and OpenAI o3 for complex tasks.

• The models differ in their capabilities, with some being more suitable for tasks like coding, math, and visual reasoning, while others excel at writing, tone control, and strategic planning.

Bottom Line Up Front

• The presentation aims to discuss how AI and OSINT are working together and how they can be used for both attacking and defending organizations

• OSINT involves collecting and analyzing publicly available information from various sources, including social media, public records, and geolocational data

• The use of AI in OSINT raises concerns about who is innovating faster, cybercriminals or cybersecurity professionals, and how to stay ahead of emerging threats

• Cyber criminals are using AI, generative AI, and agentic AI to create sophisticated attacks, including deep fakes and phishing emails

Bottom Line Up Front

• A study by Duke University researchers, including Jessica A. Reif, Richard P. Larrick, and Jack B. Soll, found that using AI tools at work can damage a person's professional reputation.

• The study, which involved over 4,400 participants, revealed that employees who use AI tools are perceived as lazier, less competent, and less diligent by their colleagues and managers.

• The social stigma against AI use was found to be consistent across different demographics, suggesting that it is a general phenomenon rather than being limited to specific groups.

• The experiments also revealed that individuals who use AI tools may be less willing to disclose their use of these tools to colleagues and managers, due to fears of being judged negatively, and that these fears are justified, as participants in the study consistently rated AI users as less competent and less diligent than non-AI users.

Bottom Line Up Front

• AI chatbots from companies like OpenAI and Google are experiencing higher "hallucination" rates, which refer to errors such as presenting false information as true or providing irrelevant answers.

• Recent testing shows that newer models, including OpenAI's o3 and o4-mini, have higher hallucination rates than previous models, with some models seeing double-digit rises in hallucination rates.

• Experts like Arvind Narayanan and Emily Bender suggest that we may have to live with error-prone AI and consider using them only for tasks where fact-checking is still faster than doing the research yourself, or avoid relying on them for factual information altogether.

Bottom Line Up Front

• A family used artificial intelligence to create a video of Christopher Pelkey, addressing his killer Gabriel Paul Horcasitas in an Arizona court.

• The AI-generated avatar was presented during Horcasitas' sentencing for shooting and killing Pelkey in a 2021 road-rage incident.

• The use of AI in the court raises ethical concerns, with law professor Harry Surden warning that it may be used to manipulate judges and juries by playing on their emotions.

Bottom Line Up Front

• As a new graduate, resilience is a critical skill to navigate the changing world of work, and it will serve as a foundation for your career.

• To succeed, it's essential to use AI in a way that complements your critical thinking skills, and to stay curious and constantly learn, as noted by experts like Aparna Chennapragada, chief product officer of experiences and devices management at Microsoft.

• Building a portfolio career, which involves a collection of roles, experiences, and side projects, can help you diversify your identity, skillset, and ability to pivot, and experts like Rosalind Chow suggest that building trust and deepening relationships by expressing gratitude and highlighting others' wins can also benefit your career.