About This Week’s Covers
This week, I used GPT to write Python code that leverages the Ideogram API and a text file to auto-generate 33 cover images in any style, downloading them directly to my local drive. This process will save an hour per week and help me catch up to the current date. The downside is that this method is single-shot image generation—there are no retakes or multiple images to choose from. It’s also much less creative, but I plan to improve it in the future. This week, I experimented by asking it to mimic Keith Haring. For the main cover, I tested three tools: MidJourney, Flux, and Ideogram. Ultimately, Ideogram won handily in terms of quality and style. The winning prompt was: “A black and white Keith Haring mural in a subway celebrating the Nobel Prize and technology, featuring iconic Haring style with curvy lines.”
Here are a few of the better API-driven ones:
This Week’s Executive Summaries
OpenAI Secures $4 Billion Credit Line, Boosting Liquidity to Over $10 Billion
OpenAI has finalized a $4 billion revolving credit line, enhancing its liquidity to more than $10 billion, alongside closing a funding round that values the company at $157 billion. The credit line, supported by financial giants including JPMorgan Chase, Goldman Sachs, and HSBC, allows OpenAI to borrow at rates of approximately 6% over three years, with an option to expand the credit to $6 billion. This financial boost follows a $6.6 billion investment from Thrive Capital, Microsoft, Nvidia, and others, aimed at expanding AI research, products, and infrastructure. Despite generating $300 million in revenue last month, OpenAI expects a $5 billion loss this year due to significant costs tied to Nvidia GPUs for training its models. The company eyes $11.6 billion in revenue by 2025 and may explore public or structured financing to sustain growth. Amid these developments, restructuring talks and leadership transitions signal evolving strategies to solidify its market position.
Cnbc | theinformation
Google DeepMind AI Scientists Awarded 2024 Nobel Prize in Chemistry
Demis Hassabis and John Jumper of Google DeepMind, along with computational biologist David Baker, have won the 2024 Nobel Prize in Chemistry for advancements in protein structure prediction and design. Hassabis and Jumper were recognized for AlphaFold, a groundbreaking AI model that predicts protein structures based on their chemical sequences, solving a decades-old challenge in biology. Baker, of the University of Washington, was honored for creating novel proteins with wide-ranging applications, including vaccine development and nanotechnology. Their work highlights AI’s transformative potential in science, offering new solutions for health and materials science.
theguardian
Two Additional AI Pioneers Awarded 2024 Nobel Prize in Physics
Geoffrey Hinton and John Hopfield, pioneers in artificial neural networks, have been awarded the 2024 Nobel Prize in Physics for their foundational work that underpins modern AI. Hopfield developed associative neural networks in 1982, enabling memory storage and pattern recognition akin to the human brain, while Hinton introduced probabilistic multilayered networks that revolutionized machine learning. Their innovations laid the groundwork for applications like language translation, facial recognition, and generative AI tools like ChatGPT. Hinton, often called the “godfather of AI,” also warned of AI’s potential risks, likening its transformative impact to the Industrial Revolution. The award highlights AI’s growing influence across disciplines, and sparked debate about its categorization as physics.
Nytimes | theguardian
Spotify’s Suno Introduces Song Editing for Pro & Premier Users
Spotify’s Suno platform now allows Pro and Premier users to edit songs by replacing lyrics or adding instrumental breaks. The feature lets users recreate sections of songs between 10 and 30 seconds. Edits can include lyric changes or prompts like “[drum break]” to add musical elements. After generating two versions of the edited section, users can select their preferred option, which integrates into the song’s overall generation process.
Sunomusic
Walmart is Testing a New AI Assistant for Shoppers and Employees
Walmart is developing its own AI assistant, Wallaby, designed to improve how customers shop and how employees work. Trained on decades of Walmart’s data, Wallaby will understand natural conversations, making it easier for customers to get help and for workers to do their jobs. Right now, it’s being tested internally, but Walmart plans to roll it out next year, starting with customer-facing tools. This new AI is part of Walmart’s broader push to make shopping and store operations smoother. From chatbots that help with returns to systems that keep shelves stocked, Walmart is betting big on AI to enhance everyday experiences.
Venturebeat
Must-See Visuals and Charts: Week Ending 10/11/2024
State of AI Report – 2024 ONLINE – Google Slides
“Two Harvard students created I-XRAY, AI glasses that can reveal personal information just by looking at someone. Details include names and home addresses. Wake-up call for privacy concerns.
“I had to test how Gen-3’s video-to-video would perform using a Viggle output. Have any of you done any interesting with vid2vid?
“Ok finally dug into Meta’s new Movie Gen paper. Text-to-video is cool and all but, to me the precise editing feature is the game changer. I mean just look at these results 🤯 It can handle complex VFX tasks like replacing environments, doing set extensions, swapping
Top 36 Links of The Week – Organized by Category
Agents and Copilots
“Sources: Sierra, an AI agent startup co-founded by Bret Taylor, is raising hundreds of millions of dollars in new funding led by Greenoaks at a $4B+ valuation (The Information)
“Awesome video by AgentOps community member @CMD_LABS showing how to build an AI songwriter with @crewAIInc @suno_ai_ and @AgentOpsAI 🖇️🖇️🖇️ (1/n)
Bret Taylor’s AI Agent Startup Nears Deal That Could Value It at Over $4 Billion — The Information
“happy to share our latest progress @rabbit_hmi on the mission building generic cross platform agent systems for everyone. here’s a sneak peak into the future – LAM desktop controlling full desktop Linux OS, just like the movie ‘her’.
“Our AI engineer can now build simple agents using the English language! Tell it what to do, and it will try to generate, execute, and deploy code. AI has already replaced SQL. The next step is to replace Python. It’s officially in its 3rd month of internship!!
Amazon Dreams of AI Agents That Do the Shopping for You | WIRED
“We’re releasing a new benchmark, MLE-bench, to measure how well AI agents perform at machine learning engineering. The benchmark consists of 75 machine learning engineering-related competitions sourced from Kaggle.
Orchestrating Agents: Routines and Handoffs | OpenAI Cookbook
“Enjoy! “Salesforce is launching a new fleet of tools for AI agents called Agentforce, and CEO Marc Benioff contends it will change the business world forever. Benioff joins Rapid Response to take us inside Salesforce’s pivot—and to slag on Microsoft’s competing CoPilot AI. He” / X
AGI (Artificial General Intelligence)
Can we make any smaller opensource LLM models smarter than human? | by Harish SG | Oct, 2024 | Medium
““Overall, we found no evidence of formal reasoning in language models (…). Their behavior is better explained by sophisticated pattern matching”” / X
“AI narrative violation detected. Severity: high. Apple’s research finds no evidence of logical reasoning in SOTA large language models like GPT-4o and o1. Needless to say, @GaryMarcus feeling good today! Popcorn worthy comments section 🍿
Anthropic
Introducing the Message Batches API \ Anthropic
“Introducing the Message Batches API—a cost-effective way to process vast amounts of queries asynchronously. You can submit batches of up to 10,000 queries at a time. Each batch is processed within 24 hours and costs 50% less than standard API calls.
“First Sam Altman’s piece on a straight path to AGI and now this piece by Anthropic’s CEO on 2 years to powerful AI, I think that any large organization or agency needs to take the possibility of near-term powerful AI seriously They may be wrong or exaggerating, but worth noting
Audio
“Introducing Director Mode. What if you could literally tell your AI voice character how to deliver a line? Now you can. After the success of Parrot Mode, we’re taking our audio studio to the next level. Our biggest release of the year is putting you in the director’s chair to
Augmented and Virtual Reality (AR/VR)
“The production value of these weather broadcasts is really next level — using unreal engine and in-camera VFX to put extreme weather events like Milton into perspective way better 2D graphics ever could
Chips, Hardware, and Infrastructure
Deep Imagination Research | NVIDIA
Ethics/Legal/Security
Northrop Grumman Adds Cutting-Edge AI Capabilities to Forward Area Air Defense | Northrop Grumman
DataTalk: A Campaign Finance Agent (Beta)
International
ByteDance’s TikTok cuts hundreds of jobs in shift towards AI content moderation | Reuters
Meta
Meta debuts new generative AI tools for creating video-based ads – SiliconANGLE
This AI Pioneer Thinks AI Is Dumber Than a Cat – WSJ
Multimodality
“The physical & digital worlds continue to collide. We’re infusing intelligence into previously ‘dumb’ sensors, like 2D video cameras. AI fuses these sensors together to extract visual and spatial intelligence.
OpenAI
OpenAI Projections Imply Losses Tripling to $14 Billion in 2026 — The Information
Perplexity
“Perplexity is hands down the best research tool on the internet right now. I *highly* recommend it if you go down rabbit holes and want to learn something fast. As someone who’s fully taught myself code, design, business, marketing, content — it truly feels like a life hack. I” / X
“I asked the CEO of Perplexity if we’ll ever see a money-making machine like Google again 🤑 Plus, the four ways Perplexity is generating revenue and its plans to collaborate with publishers and developers. Listen to my full episode with @AravSrinivas on The TED AI Show 🎙️
Publishing News: Week Ending 10/11/2024
“I asked the CEO of Perplexity if we’ll ever see a money-making machine like Google again 🤑 Plus, the four ways Perplexity is generating revenue and its plans to collaborate with publishers and developers. Listen to my full episode with @AravSrinivas on The TED AI Show 🎙️
Meta debuts new generative AI tools for creating video-based ads – SiliconANGLE
OpenAI partners with Cosmopolitan and Elle publisher Hearst
OpenAI and Hearst Content Partnership | OpenAI
“BREAKING: Looks like OpenAI is entering the arena against Perplexity… citations are now in GPT-4o 👀
Robotics and Embodiment
This Homemade Drone Software Finds People When Search and Rescue Teams Can’t | WIRED
Video
“Ok finally dug into Meta’s new Movie Gen paper. Text-to-video is cool and all but, to me the precise editing feature is the game changer. I mean just look at these results 🤯 It can handle complex VFX tasks like replacing environments, doing set extensions, swapping
Meta debuts new generative AI tools for creating video-based ads – SiliconANGLE
“Look at the realism here! Hot take: This blows most ‘old-school’ CG humans out of the water. 😮 “But you have no control”.. it’s getting there! Using ‘Facepoke’ as input for Runway’s video-to-video, it’s clear we’re entering a new era of creation. #AIArt #GenerativeAI https://x.com/MartinNebelong/status/1843409578438107519
Leave a Reply