AI Accelerates, Battles Intensify, Breakthroughs Reshape the Future

Grab your favorite beverage and get ready for an edition of your (hopefully) favorite weekend read on all things future, innovation, and disruption, brimming with captivating content.

Mar 22, 2024

Estimates of effective compute doubling from algorithmic improvements across different domains.

”The Report of My Death Was an Exaggeration.” The famous Mark Twain quote echoes through the world of AI, where some pundits bemoan the fact that OpenAI’s flagship model, GPT-4, hasn’t seen a significant update since its launch a year ago—and no other model has superseded it in a substantial way since. However, a recent study provides compelling evidence of the rapid pace of LLM development: LLMs are improving several times faster than Moore’s Law, the original engine of the digital age. For LLMs, the primary measure in the paper is model performance, and the authors find that the compute needed to achieve a given level of AI ability is halving every 5 to 14 months, as opposed to 24 months for Moore’s Law. Moreover, the paper demonstrates that the speed of improvement is not slowing down over time—at least not for now. Buckle up—AI is alive and thriving.

GPT-2, Once Declared “Too Dangerous” for Public Release, Now Runs Inside an Excel Spreadsheet. The race to run large language models on smaller and smaller hardware has been on for a while—we have seen plenty of models being released that are small enough to run on your laptop yet perform at the level of GPT-3 to 3.5. Now, a researcher has implemented a working version of an LLM that is about as capable as GPT-2 inside an Excel spreadsheet as a way for people to learn how LLMs work. Prepare for LLMs running on less and less capable hardware—and thus becoming truly ubiquitous.

Alt-Meat Goes on the Offensive: After meat producers waged war on meat alternatives (arguing that they’re not “meat”—the same argument being brought forward against the alt-milk industry), alt-meat giant Impossible declared war on the butcher aisle. The company is rolling out a bold, bloody-red package redesign that ups the ante. Let the battle commence! (and here is an interest statistic: In the US more than 20% of all milk sales are now alt-milks)

Palmer Luckey, Founder of the Ur-VR Company Oculus, Wages War: Luckey, the visionary behind the groundbreaking Oculus VR headset, has set his sights on the battlefield with his company, Anduril Industries. Their latest creation, the ALTIUS-700M drone, is a game-changer in military technology. Powered by the AI-driven Lattice system, these drones act autonomously, circling the skies until they lock onto their targets with devastating precision. The 700M’s impressive range and firepower, comparable to the costly AGM-114 Hellfire missile, promise to redefine modern warfare. The autonomous era has arrived, and it’s reshaping the future of conflict—surely not without its own set of challenges (technological and ethical).

Talking about Luckey, Oculus, and VR: Apple’s Vision Pro: A Devkit in Disguise? Hugo Barra, former Head of Oculus at Meta, takes us on a deep dive into Apple’s newly unveiled Vision Pro headset, dissecting its hardware and software with surgical precision. While the device bleeds genius and audacity, Barra argues it’s more of an over-engineered “devkit” than a product ready for mass adoption—but one that could finally give Meta its Android moment.

Gene-editing Breakthrough Leads to Pig Liver Transplant. If you have ever witnessed one of our Disruption Mapping exercises, you will have heard us discuss the seemingly peculiar implication of widespread autonomous vehicle adoption on organ donation—assuming self-driving cars result in fewer fatalities, we will inevitably have (even) fewer organ donors. In our sessions, we often mention this as a call for innovative ways to supply organs to those in need. Now, a Chinese patient has received the first gene-edited pig liver transplant.

Outsmarting the Future, Together. Samotsvety, a group of skilled forecasters, consistently outperforms experts and intelligence agencies in predicting world events. Their success lies in quantitative reasoning, constant practice, and a keen understanding of base rates and patterns in global affairs, allowing them to build accurate models of how the world works—and occasionally, how it might change.

What We Are Reading

🌱 The Benefits of Innovation That Isn’t Disruptive Innovation and growth don’t always require disruption. Surprising, isn’t it? Another approach is to create something new beyond existing industry boundaries, offering unique organizational and business advantages. @Jane

🌊 Many Coral Reefs Are Dying. This One Is Exploding with Life. Have you ever wondered how corals reproduce? It’s called spawning, and this mesmerizing video is worth watching to see it in action. @Mafe

🤷‍♂️ Of Course America Fell for Liquid Death Nothing captures the enduring power of a brand like the preposterous rise of Liquid Death. @Jeffrey

🙅‍♂️ Every Location Has Got Worse for Getting Actual Work Done Slightly critical in tone, but it touches on many points that most will know and recognize. It does show how abundant distractions are and how intentional space selection has become quite a decisive factor for work. @Julian

🤐 How to Spot “Pluralistic Ignorance” before It Derails Your Team Pluralistic ignorance occurs when people stay silent about their doubts or concerns because they assume everyone else agrees with a certain belief or course of action. @Pedro

🔄 Everyone’s a Sellout Now In a world where everyone is a brand, artists face the tyranny of self-promotion, sacrificing creativity for clicks in the digital rat race. @Pascal

Bits & Pixels

» Neo-luddites are a thing, and they are not particularly optimistic about our technology-driven future.

» On the other end of the spectrum, techno-optimist Ray Kurzweil finds himself in a (in places extremely weird) conversation with Joe Rogan about his love for suspenders and why AI will rule us all.

» The solar revolution is depleting the world’s groundwater—a stark reminder that we often encounter unforeseen repercussions when it comes to novel technologies and their widespread adoption.

» And again, on the other side, here are Michael Liebreich’s, Senior Contributor at BloombergNEF, arguments on why Net Zero is achievable—and the path we must take to reach it.

» Remember our radical Weekend Briefing last week? AutoDev introduces fully automated, AI-driven software development to your IDE, emerging as another contender in what appears to be a rapidly expanding market. And QuietSTaR teaches LLMs to think before they speak.

» Zoom has never looked more uncanny: Google’s VLOGGER takes a single picture and transforms it into a fully animated avatar.

» The future of marketing could well be Paris Hilton and Roblox. The hotel heiress, who not only has a successful DJ career but also runs a media consultancy, published a report highlighting brand opportunities in the virtual world of Roblox, where her own branded experience, Slivingland, drew 3.4 million visitors.

» Still confused about how Large Language Models actually work? Here’s a simple explainer video to clear things up.

» Venture capital powerhouse Andreessen Horowitz’s Top 100 GenAI Consumer Apps report is consistently worth perusing.

» Here’s (another) good AI resource—Let’s Build AI.

» Will AI replace software developers? Not so fast, argues Sheshbabu Chinnakonda.

» The next AI superpower? Denmark!

» Thanks to improved models, (much) more data, and massively increased computing power, four-day weather forecasts are now as accurate as one-day forecasts were 30 years ago.

» Nvidia’s boss has some stern words for Stanford graduates: “I wish upon you ample doses of pain and suffering.”

» When it comes to climate change, never underestimate the power of Mother Nature: a $500,000 sand dune, built to protect US homes, vanished in mere days.

Candy 🍬

The fine folks at lumiQ recorded one of their CPE-credit worthy sessions with yours truly on “anticipating market changes”.

🎧 A Futurist’s Guide to Anticipating Market Changes

And the amazing team at the Better Learning Podcast / Second Class Foundation talked to me about ChatGPT & Disruption in the Classroom.

🎤 Pascal Finette | ChatGPT & Disruption in the Classroom

Some Fun Stuff

Somewhat morbid, yet a poignant reminder to cherish life: Discover this beautifully designed “Estimate Your Age of Death” calculator. Surprisingly, my chosen home (USA) has me passing away (on average) two years earlier than my birth country (Germany). 🪦🌈🌞

Some random fun links: