Hello Tuners,

OpenAI is finally entering the search arena with its shiny new ChatGPT Search feature, leaving some to wonder if they’re really innovators or just playing catch-up with Google and Perplexity. Speaking of Perplexity, its CEO is juggling accusations of “content kleptocracy” like a pro, dodging questions about plagiarism while promising to play nice with publishers; good luck!

Meanwhile, Ashton Kutcher is shaking up the AI scene with a hefty investment in World Labs, aiming to create immersive experiences that could make Hollywood sit up and take notice. And over at GitHub, they're rolling out updates that make coding as easy as pie, blending multiple AI models and adding new features to put the “fun” back in functional programming. So, buckle up as we dive into these stories that are as entertaining as they are enlightening!

• OpenAI’s New Search Feature Enters a Crowded Arena

• “Content Kleptocracy”? Fair Use or Free Riding?

• Ashton Kutcher, Who Told You About AI?

• GitHub Shakes Hands with Apple’s Xcode

OpenAI’s New Search Feature Enters a Crowded Arena

OpenAI has finally rolled out ChatGPT Search, apparently deciding it’s time to join the game that Google and Perplexity have been playing since, well… forever. After two years of AI domination (debatable), OpenAI’s big move is a “timely answers” search feature. You’d almost think they invented it, though it’s suspiciously similar to what’s been sitting over on Google’s and Perplexity’s desks for ages. While Perplexity’s CEO is still stuck in court trying to prove they didn’t raid the internet’s cookie jar for news, OpenAI’s conveniently glossing over that they’ve also been accused of the same thing. They’re promising attribution, in-line links, and responsible sourcing.

OpenAI also adds that they’re “listening to publishers” to avoid eating into their traffic too much because nothing says “we’re on your side,” like taking your content and summarizing it (try asking GPT4o to summarize a research paper by the way), and serving it to users so they don’t have to visit your site. And the kicker? OpenAI made a browser extension to make ChatGPT the default Chrome search engine, as if we all haven’t figured out how to set Google as our homepage.

“Content Kleptocracy”? Fair Use or Free Riding?

Perplexity’s CEO, Aravind Srinivas, isn’t entirely clear on scraping content, a stance made even muddier at his recent TechCrunch Disrupt interview. News Corp, among others, has dubbed Perplexity’s approach a “content kleptocracy” and sent cease-and-desist letters to hammer home that using others’ content, summarizing or not, without proper rights isn’t fair play. When Srinivas dodged defining “plagiarism” outright, he argued that Perplexity simply “surfaces” information like any academic or journalist, conveniently leaving out that paraphrasing 48% of a Forbes article looks more like rebranding than re-reporting.

The public relations spin continued with Srinivas promising partnerships with publishers, while lawsuits from Dow Jones argue that Perplexity cannibalizes traffic by summarizing without linking back. This, critics say, draws users away from sources that bear the brunt of production costs. Meanwhile, Perplexity’s lofty pitch about “universally distributing” facts doesn’t quite square with how its $8 billion valuation is built on borrowed work, leaving actual news creators wondering if there’s any room for fair compensation in Perplexity’s brave new world.

Ashton Kutcher, Who Told You About AI?

Ashton Kutcher’s venture capital firm, Sound Ventures, is diving into the AI landscape with a significant investment in Fei-Fei Li’s World Labs, a startup focused on developing advanced "large world models" for the 3D environment. Announced at TechCrunch Disrupt 2024, this investment is part of a larger $265 million AI fund that has previously backed notable companies such as OpenAI and Anthropic. With World Labs valued at over $1 billion after raising $230 million from various investors, including a16z and NEA, the firm aims to cater to game developers and movie studios. In this area, Kutcher's Hollywood background may provide valuable insights.

Sound Ventures also explores innovative AI hardware and software combinations, suggesting that existing form factors and operating systems may not fully optimize AI technology. General partners Guy Oseary and Effie Epstein highlighted their interest in startups that redefine AI's physical presence, mentioning their past interactions with companies like Humane and Rabbit. Additionally, Kutcher revealed ongoing discussions with legendary designer Jony Ive about a new AI device in collaboration with OpenAI CEO Sam Altman, hinting at the potential for groundbreaking advancements in the industry.

GitHub Shakes Hands with Apple’s Xcode

At the GitHub Universe conference, GitHub unveiled significant updates to its AI-powered development tools, including GitHub Copilot. Integrating multiple AI models, such as Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s GPT-4o variants, allows developers greater flexibility in choosing the best model for their coding tasks. GitHub's enhanced partnership with Microsoft also brings multi-file editing capabilities to its VS Code IDE, streamlining the development process. Additionally, GitHub is expanding Copilot's availability to Apple’s Xcode and launching a new Stack Overflow extension to provide developers with instant insights from the community.

Moreover, GitHub introduced the Spark initiative, which aims to democratize software development by allowing non-developers to create applications quickly. The GitHub Copilot Workspaces feature receives updates to enhance collaboration and code review, enabling AI-assisted pull requests tailored to team configurations. With these advancements, GitHub aims to unlock the potential for one billion developers by 2030, emphasizing accessibility and ease of use in software creation.

Weekly Research Spotlight 🔍

Agentic Information Retrieval

The future of information retrieval (IR) is poised for significant evolution, driven by advancements in large language models (LLMs) and the emergence of a new paradigm known as Agentic Information Retrieval (Agentic IR). Traditional IR systems, which have existed since the 1970s, rely on predefined architectures to filter and retrieve information based on user queries. Despite their success, these systems have task flexibility and interactivity limitations, often requiring users to refine searches iteratively. The introduction of LLMs has opened up new possibilities for IR, allowing for complex reasoning and multi-step interactions through AI agents. Agentic IR aims to redefine the retrieval process by enabling agents to understand user intent and engage in dynamic interactions, broadening the scope of tasks they can handle.

Agentic IR distinguishes itself from conventional systems by employing a unified architecture where AI agents utilize observation, reasoning, and action in a recurring manner rather than operating within a single interaction. This new framework enhances the agent's capabilities through prompt engineering, retrieval-augmented generation, and reinforcement learning, enabling them to adapt and optimize their performance over time. With applications ranging from life assistants to coding support, Agentic IR represents a transformative shift in how users access and interact with information, potentially becoming the central entry point in future digital ecosystems. As these technologies advance, they promise to enhance user experience and create innovative applications that leverage the full potential of AI in information retrieval.

LLM Of The Week

Aya Expanse

Cohere For AI has unveiled Aya Expanse, a groundbreaking family of multilingual models optimized for performance across 23 languages. Available in 8 billion and 32 billion parameters, these models can be accessed on Kaggle and Hugging Face, showcasing Cohere's commitment to advancing multilingual AI. The 32B model sets a new standard, outperforming notable competitors like Gemma 2 27B and Llama 3.1 70B, while the 8B version leads its parameter class, achieving impressive win rates against peers.

Key innovations driving the success of Aya Expanse include a novel data arbitrage strategy for synthetic data generation, ensuring quality outputs in low-resource languages, and enhanced preference training incorporating diverse cultural perspectives, improving both performance and safety. Additionally, model merging techniques have been employed to combine strengths from various models, contributing to the state-of-the-art results in multilingual capabilities. This release marks a significant milestone in bridging the language gap and improving AI's effectiveness globally.

Best Prompt of the Week 🎨

Today's Goal: Try new things 🧪

Acting as an Education Counseling Planner

Prompt: I want you to act as an educational counseling planner. You will create a structured daily plan specifically designed to help a study abroad consultant effectively guide students who visit their office in selecting the most suitable courses and provide comprehensive educational counseling. You will identify key steps for assessing student needs, develop strategies for course selection and career guidance, select resources to streamline consultations, and outline additional activities to enhance the counseling experience. My first suggestion request is: "I need help creating a daily activity plan for a study abroad consultant who assists students in choosing the best courses and provides educational counseling."

This Week’s Must-Watch Gem 💎

This Week's Must Read Gem 💎