The coming wave of AI PCs; inside the AI research boom; high valuations are hurting AI startups in the race for talent; the future of gen AI apps is in the enterprise
Some AI startups struggle to find market fit; Ethan Mollick's profile in the WSJ; Abu Dhabi hosts the first Autonomous Racing League; the FT tests OpenAI's Sora model; China's four OpenAI competitors
For years, heavy-duty AI workloads were confined only to the data centers of big tech companies and specialized research labs. But a new category of personal computing device is about to bring advanced AI capabilities into homes and offices around the world: the AI PC.
While Microsoft has recently tried to define them using hardware specs (and by adding a new key to Windows keyboards!), AI PCs are essentially desktop computers or laptops with added hardware that’s capable of large model inference (and perhaps even training). Unlike traditional PCs which rely mostly on CPUs for general computing tasks, an AI PC has a more heterogeneous compute architecture that allows it to lean heavily on modern GPUs or other specialized AI accelerators (known as NPUs) to handle the hundreds of trillions operations per second (TOPS) required for large AI workloads.
Several leaders in computing hardware are already developing AI PCs, with expected product launches later this year:
NVIDIA recently unveiled the GeForce RTX SUPER desktop GPUs used by manufacturers including Acer, ASUS, Dell, HP, Lenovo, MSI, Razer and Samsung for upcoming AI laptops that will deliver TOPS-related performance increases of 20x to 60x compared with previous generation systems.
Qualcomm has announced a series of Snapdragon X Plus and Snapdragon X Elite chips for AI PCs expected to ship from OEMs starting mid-2024. The Snapdragon X lineup is notable because it has been optimized for an upcoming version of Windows designed to natively run AI software on the Arm CPU architecture.
Apple's next-generation Macs are expected to pack a custom Apple Silicon chip designed from the ground up for AI and machine learning tasks.
So what could this new supercomputing power offer to the average person just a few months from now?
Here are a few possibilities based on demonstrations we’ve seen so far:
Hyper-intelligent assistants: Windows was the first operating system to integrate a more advanced AI assistant in the form of Copilot. However, Copilot runs in the cloud and cannot answer basic queries and handle simple tasks when it’s not connected to the internet. But the large language models running directly on an AI PC could provide a digitally intelligent assistant that people can use on the go and will be almost as capable as a human expert across nearly any domain: writing, analysis, coding, math, creative projects and more. Need an essay ghostwritten or lines of code debugged in an obscure programming language? Your AI assistant will be up to the task.
Real-time multimedia intelligence: Today's AI models can already automatically caption images, transcribe audio and video, and generate synthetic media. But an AI PC could take these capabilities to new heights, instantly creating high-quality keynote presentations from a simple prompt or deliver personalized media experiences (such as music or movies)—all in real-time as the content is consumed.
Photorealistic synthetic media: One of the most interesting capabilities of modern AI is its ability to generate incredibly realistic synthetic images, video, audio, and text. While the industry has made huge progress in the last 12-18 months, current methods are still limited because they require access to cloud computing. But an AI PC could open the floodgates for creators to leverage vast AI models to generate photorealistic digital content on demand for videos, games, marketing, and more.
Even though I’ve been skeptical of the utility offered by mobile AI computing devices, the coming wave of AI PCs has the potential to create a broad range of new uses cases, creating a similar effect to the original PC revolution enabled by the transition to personal computing in the 1980s and 1990s.
And now, here are this week’s news:
❤️Computer loves
Our top news picks for the week - your essential reading from the world of AI
The Guardian: ‘Eugenics on steroids’: the toxic and contested legacy of Oxford’s Future of Humanity Institute
FT: The Financial Times and OpenAI strike content licensing deal
WSJ: AI Startups Have Plenty of Cash. They Often Don’t Yet Have a Business.
WSJ: Meet the AI Expert Advising the White House, JPMorgan, Google and the Rest of Corporate America
Axios: Inside the AI research boom
The Verge: In the first Autonomous Racing League race, the struggle was real
Wired: Recruiters Are Going Analog to Fight the AI Application Overload
FT: How good is OpenAI’s Sora video model — and will it transform jobs?
Wired: The Unsexy Future of Generative AI Is Enterprise Apps
FT: Four start-ups lead China’s race to match OpenAI’s ChatGPT
⚙️Computer does
AI in the wild: how artificial intelligence is used across industry, from the internet, social media, and retail to transportation, healthcare, banking, and more
The Telegraph: AI-backed app could help elderly and save £1bn a year
VentureBeat: BioRaptor and Aleph Farms use AI to lower the costs of cultivated beef
Bloomberg: AI Is Helping Automate One of the World’s Most Gruesome Jobs
TechCrunch: Airbnb releases group booking features as it taps into AI for customer service
Wired: A Vast New Data Set Could Supercharge the AI Hunt for Crypto Money Laundering
TechCrunch: Yelp is launching a new AI assistant to help you connect with businesses
The Telegraph: How AI could become an unlikely ally in the fight against forest fires
Fortune: Japan has had so many bear attacks in the past year it’s turning to AI to act as a warning system
Wired: I Tried These AI-Based Productivity Tools. Here’s What Happened
Business Insider: I'm using my synesthesia to create a new genre of AI art. The technology can 'read' my paintings and help me compose music.
The Atlantic: I Witnessed the Future of AI, and It’s a Broken Toy
Washington Post: An AI blood test purports to diagnose postpartum depression
The Information: How Tesla Used a San Francisco Neighborhood to Improve Its Driverless Technology
The Verge: ChatGPT’s AI ‘memory’ can remember the preferences of paying customers
BBC: AI technology in Northamptonshire catches drivers on mobiles
🧑🎓Computer learns
Interesting trends and developments from various AI fields, companies and people
WSJ: Online Marketplaces Like eBay, Etsy Are Counting on AI to Supercharge Shopping
Bloomberg: JPMorgan Unveils IndexGPT in Next Wall Street Bid to Tap AI Boom
Fortune: Runway’s second-ever AI Film Festival walks the line between movie business’ past and its future
CNBC: AI engineers report burnout, rushed rollouts as ‘rat race’ to stay competitive hits tech industry
Fortune: Tech leaders crave ‘peace of mind’ with AI in the cloud
Wired: Nick Bostrom Made the World Fear AI. Now He Asks: What if It Fixes Everything?
FT: Wanted: a data standard to underpin lawyers’ use of generative AI
WSJ: The AI-Generated Population Is Here, and They’re Ready to Work
Fortune: Employers toggle the deployment of AI for their workforce and customers
WSJ: These Models Gave Up Photoshoots to Sell Their AI Likenesses
VentureBeat: The first music video generated with OpenAI's unreleased Sora model is here
VentureBeat: Vercel acquires ModelFusion, launches AI SDK 3.1 for enterprise AI development
AP: AI use by businesses is small but growing rapidly, led by IT sector and firms in Colorado and DC
Business Insider: Biden used ChatGPT for the first time. Here's how that went.
Semafor: ‘Disgorgement’: Amazon researchers suggest ways to get rid of bad AI data
Android Authority: Rabbit R1, a thing that should just be an app, actually is just an Android app
New York Times: From Baby Talk to Baby A.I.
WSJ: The Last Stock Photographers Await Their Fate Under Generative AI
VentureBeat: Ideogram launches Pro tier with 12,000 fast-AI image generations per month
VentureBeat: New AI-powered Iterable features help brands cut through the noise
VentureBeat: Atlassian introduces Rovo, an AI-powered knowledge discovery tool for the enterprise
VentureBeat: Nvidia launches tech update for ChatRTX AI chat
VentureBeat: Mindtrip’s AI travel assistant aims to be your one-stop shop for trip planning
VentureBeat: Amazon Q enterprise AI chatbot is now generally available
Reuters: Anthropic releases business chatbot in hunt for corporate dollars
FT: Apple targets Google staff to build artificial intelligence team
MIT Technology Review: Sam Altman says helpful agents are poised to become AI’s killer function
Business Insider: Google says immigration rules are making it hard to hire top AI talent
The Telegraph: Salman Rushdie has conversation with AI version of his attacker
Business Insider: CEO of Anthropic — the AI company Amazon is betting billions on — says it could cost $10 billion to train AI in 2 years
TechCrunch: How RPA vendors aim to remain relevant in a world of AI agents
Bloomberg: Apple Intensifies Talks With OpenAI for iPhone Generative AI Features
CBS 60 Minutes: Nvidia CEO Jensen Huang and the $2 trillion company powering today's AI
ZDNet: Apple builds a slimmed-down AI model using Stanford, Google innovations
Business Insider: AI has yet to be truly transformational like the internet, CEO says
TechCrunch: Curio raises funds for Rio, an ‘AI news anchor’ in an app
Bloomberg: AI Boom’s Secret Winners? The Companies Expected to Power It
Business Insider: AI is making managers nervous
Axios: Musk gets self-driving Teslas tentatively cleared in China during surprise trip
Business Insider: A centuries-old Japanese ceramics maker has found itself at the center of the AI revolution
Fortune: Fears over AI killing labor demand are ‘probably overblown,’ researcher argues in new Google report
FT: The Algorithm by Hilke Schellmann — why AI really is coming for your job
VentureBeat: GitHub previews Copilot Workspace, an AI developer environment to turn ideas into software
TechCrunch: Creators of Sora-powered short explain AI-generated video’s strengths and limitations
VentureBeat: Microsoft deputy CISO says gen AI can give organizations the upper hand
VentureBeat: Google’s DeepMind creates ‘Gecko’, a rigorous new standard for testing AI image generators
Keep reading with a 7-day free trial
Subscribe to Computerspeak by Alexandru Voica to keep reading this post and get 7 days of free access to the full post archives.