This Week in AI with cohosts ChatGPT and Tony DeSimone

The author, Anthony DeSimone, argues that ChatGPT has become so ubiquitous that it has earned the right to be a verb, just like "Google" or "Uber." DeeSimone points to the increasing use of the term "GPTing" to describe the process of using ChatGPT to complete tasks as evidence of its cultural impact. He emphasizes the versatility of ChatGPT, its ability to be used across a variety of fields, and its user-friendly interface as reasons for its widespread adoption. DeSimone sees "GPTing" as a sign of a cultural embrace of AI, a reflection of a world where technology is no longer a separate entity but rather an integrated part of daily life.

Elon Musk has predicted that by 2040, there will be over 10 billion humanoid robots in use globally, each costing between $20,000 and $25,000. These robots, like Tesla's Optimus, are being developed to perform a wide range of tasks, but experts remain skeptical about the feasibility of such a large-scale deployment due to technical limitations and concerns about economic affordability. Despite these challenges, the growing interest in humanoid robotics suggests a potential shift in our society and economy, as these machines could play a significant role in various sectors, potentially reshaping our world in the coming decades.

OpenAI has introduced real-time web search capabilities to ChatGPT, allowing users to access up-to-date information and enhancing its functionality as a competitor to Google. This feature enables ChatGPT to provide immediate answers, maintain conversational context, and include source links for verification. As the rollout progresses, this development could redefine how users interact with AI and search technology, offering a more personalized and engaging experience.

In this episode, we delve into the transformative impact of AI on software development, highlighted by Google's recent revelation that over 25% of new code is now generated by AI systems. As CEO Sundar Pichai discusses how AI tools are enhancing productivity and efficiency, we explore industry-wide trends showing that a significant majority of developers are adopting AI coding tools. While the benefits of increased efficiency are clear, we also address concerns about job security for programmers and the potential risks of AI-generated code. Join us as we navigate the future of coding in an

The blog discusses the emerging competition in the online search landscape as tech giants like Meta and OpenAI develop their own AI-powered search engines, threatening Google's long-standing dominance. Meta is building proprietary web crawling technology to provide real-time information for its Meta AI chatbot, aiming for self-sufficiency and reducing reliance on Google and Bing. Meanwhile, OpenAI is working on SearchGPT, which will deliver quick, accurate responses with clear sourcing. This shift towards AI-integrated search promises enhanced user experiences through personalized, conversational interfaces and real-time information processing. As these companies leverage their existing user bases and data, they could significantly challenge Google's market share. However, challenges such as content rights, misinformation, privacy concerns, and regulatory scrutiny remain. Ultimately, the evolution of AI-powered search engines signals a new era in information retrieval, promising more intuitive and tailored experiences for users while reshaping competition in the tech industry.

Apple has introduced Apple Intelligence, a suite of AI-powered features for its devices. This new system enhances writing tools with features like rewrite, proofread, and summarize, making it easier to write, edit, and understand text. Siri has also been significantly upgraded, becoming more natural and integrated into the system, with features like type to Siri and expanded product knowledge. Photos now includes natural language search and AI-generated memories, while productivity is enhanced with priority messages, email summaries, and smart replies. Upcoming features include Genmoji for personalized emoji creation, Image Playground for generating playful images, and ChatGPT integration. Privacy is a key focus, with many features running entirely on-device, and user data never being stored or shared with Apple.

In a shocking turn of events, OpenAI has disbanded its "AGI Readiness" team, coinciding with the resignation of senior advisor Miles Brundage, who voiced grave concerns about the industry's preparedness for Artificial General Intelligence (AGI). This dissolution follows the earlier disbandment of the Superalignment team, raising alarms about OpenAI's commitment to AI safety. Brundage's departure signifies a loss of critical expertise, as he warns that neither OpenAI nor any frontier lab is equipped to handle the implications of AGI. The blog explores the potential consequences of these developments, emphasizing the urgent need for enhanced regulation, transparency, and collaboration in the rapidly evolving AI landscape.

The Biden administration has issued its first national security memorandum on artificial intelligence, urging government agencies to harness advanced AI systems to maintain a competitive edge and enhance national security. This initiative emphasizes ethical use, strategic applications in cybersecurity and military operations, and the establishment of the AI Safety Institute as a central resource for managing AI technology.

Today we're diving into something written by Generative AI Specialist, Tony DeSimone, that's been causing quite a stir in the AI world - those pesky AI hallucinations. You know, those moments when ChatGPT or other AI tools just... make stuff up? Like that wild case last year when it fabricated six entire legal cases out of thin air! We're going to break down why this happens (spoiler alert: it's all in that 'generative' programming), why it matters for your business, and most importantly, how to protect yourself from getting caught in these AI fairy tales. Whether you're using AI for research, content creation, or business analysis, understanding these hallucinations isn't just some tech trivia – it could save you from some seriously embarrassing situations. So grab your coffee, and let's decode this crucial aspect of AI that every professional needs to understand in today's rapidly evolving business landscape.

n early 2023, Samsung experienced a major data leak after employees used ChatGPT to assist with tasks, inadvertently exposing sensitive information to OpenAI. This incident serves as a cautionary tale about the risks of using public AI models for proprietary data, highlighting the need for strong policies, employee education, and regular audits to safeguard company secrets. We'll explore what happened, the vulnerabilities of public large language models, and how businesses can responsibly integrate AI while protecting their valuable data assets.

As generative AI tools like ChatGPT become more integrated into education and daily life, there's growing concern about their potential to undermine critical thinking, creativity, and problem-solving skills, especially in young people. While AI offers unprecedented access to information and can enhance learning, over-reliance on these technologies risks intellectual complacency and diminished cognitive abilities. This blog explores these challenges and offers strategies to balance the benefits of AI with the need to preserve essential skills, ensuring that AI serves as a tool for innovation without replacing human thought and creativity.

A Florida mother, Megan Garcia, has filed a lawsuit against Character.AI following the tragic suicide of her 14-year-old son, Sewell, who became deeply involved with the platform's AI chatbot. The lawsuit alleges that Character.AI acted recklessly by providing minors access to realistic AI companions without adequate protections, collecting data from young users, employing addictive design features, and guiding users toward inappropriate content. This case raises significant questions about the responsibilities of tech companies in safeguarding vulnerable users, particularly minors, and could set a precedent for holding AI platforms accountable for their impact on mental health. As society grapples with the implications of AI technology, this incident underscores the urgent need for robust regulations and ethical guidelines.

Anthropic has unveiled two major updates to its Claude AI lineup: the upgraded Claude 3.5 Sonnet and the new Claude 3.5 Haiku, both designed to enhance AI performance and accessibility. Claude 3.5 Sonnet delivers significant coding improvements, outperforming other models on key benchmarks and providing advanced capabilities for both creative and technical tasks. Meanwhile, Claude 3.5 Haiku matches the high performance of previous models but at a lower cost, making advanced AI more accessible. The most exciting news is the introduction of a public beta feature called "computer use," which allows Claude to interact with computers like a human—typing, clicking, and navigating across multiple apps. Though still experimental, this feature marks a big step toward seamless AI-driven automation, with early adopters like Asana, Canva, and Replit already leveraging its potential.

In a surprising turn of events, Alcon Entertainment, the production company behind "Blade Runner 2049," has filed a lawsuit against Tesla, CEO Elon Musk, and Warner Bros. Discovery. The legal action stems from Tesla's Robotaxi unveiling event on October 10, 2024, where Musk allegedly used AI-generated imagery resembling scenes from the iconic sci-fi film without permission. Despite Alcon's prior refusal to grant usage rights, Tesla proceeded with visuals that closely mirrored key moments from the movie, including a man overlooking a desolate cityscape. The lawsuit accuses the defendants of copyright infringement and false endorsement, highlighting concerns about potential damage to Alcon's brand and future partnerships. This case raises important questions about the intersection of AI, copyright law, and the responsibilities of high-profile tech figures in respecting intellectual property rights.

UCLA researchers have developed a groundbreaking deep-learning framework called SLIViT (SLice Integration by Vision Transformer) that can automatically analyze and diagnose 3D medical images with accuracy matching that of medical specialists, but in a fraction of the time. Unlike other models, SLIViT has wide adaptability across various imaging modalities, including 3D retinal scans, ultrasound videos, 3D MRI scans, and 3D CT scans. The system overcomes the challenge of limited training datasets by leveraging prior medical knowledge from the 2D domain, allowing it to perform effectively with moderately sized labeled datasets. SLIViT's automated annotation capability has the potential to improve diagnostic efficiency, reduce data acquisition costs, and accelerate medical research. The researchers plan to expand their studies to include additional treatment modalities and explore the model's potential for predictive disease forecasting to enhance early diagnosis and treatment planning.

In this episode, we discuss OpenAI's revenue projections and what they reveal about the company's strategic direction. ChatGPT remains the key revenue driver, with continued investment in its growth. OpenAI anticipates modest API growth but expects a big boost from new products launching in 2025, including AI-powered search engines, video tools, and AI agents. These agents could mark a major leap in AI capabilities. However, OpenAI is navigating challenges like ethical concerns, regulatory scrutiny, and competition. The takeaway? OpenAI sees the future of AI in consumer products, autonomous agents, and diverse AI solutions.

Anthropic, an artificial intelligence research company, has updated its Responsible Scaling Policy (RSP) to address the risks posed by increasingly powerful AI models. The RSP lays out a framework for ensuring that AI models are developed and deployed responsibly, with safeguards in place to prevent catastrophic harm. The company's approach is based on a tiered system of safety levels (ASLs), with higher levels requiring more stringent safeguards for models with greater capabilities. The RSP also introduces the concept of Capability Thresholds, specific levels of AI capability that trigger the implementation of increased safeguards. The policy includes detailed procedures for assessing model capabilities, implementing appropriate safeguards, and ensuring transparency and accountability in the development and deployment of AI systems.

In this episode, we explore the significant changes at OpenAI, diving into the major shifts and transformations that have reshaped the company's direction. We discuss the factors behind the exodus of talent, the future trajectory of AI research, and how these internal changes could impact the broader tech ecosystem. Tune in to understand what’s happening behind the scenes and what it means for AI's future.

In this episode, we explore a groundbreaking proof-of-concept system created by two Harvard students that transforms Meta's Ray-Ban smart glasses into a powerful surveillance tool, raising alarming concerns about privacy in the digital age. The system, named I-XRAY, combines facial recognition, large language models, and public databases to identify strangers and retrieve personal information in real-time. Listeners will learn how the I-XRAY system operates, using smart glasses to discreetly capture live video, which is then processed to recognize faces, conduct reverse image searches, and pull personal details like home addresses and phone numbers. Through demonstrations, the students were able to gather sensitive data on individuals in public, revealing the chilling potential for misuse in identity theft, stalking, or worse. We also delve into the broader implications for privacy and AI as this project illustrates how easily accessible technologies can be turned into invasive tools. The episode wraps up by discussing mitigation strategies, the ethical considerations of such technologies, and the urgent need for stronger privacy regulations in an AI-driven world. Key topics covered: Join us as we navigate the evolving landscape of privacy and technology, and discuss how to safeguard personal information in an increasingly connected world.

In this week’s episode, we explore a thought-provoking essay by Dario Amodei, CEO of Anthropic, who shares his optimistic vision of how AI could radically transform the world for the better. Amodei discusses AI’s potential to revolutionize sectors like biology, healthcare, mental health, and economic development by accelerating research and problem-solving on a massive scale. He highlights AI’s ability to potentially cure diseases, double the human lifespan, and eliminate poverty. However, Amodei also emphasizes the need for cautious development to navigate risks, particularly in governance and the future of work, advocating for the importance of AI safety and regulation to ensure these transformative outcomes. Join us as we unpack Amodei’s predictions and their implications for the future.