🔮 Emancipated robots; DIY superconductors; Sinoamerican disputes; neotinder ++ #434
Your insider guide to AI and exponential technologies
Hi, I’m Azeem Azhar, founder of Exponential View. As a global expert on exponential technologies, I advise governments, some of the world’s largest firms, and investors on how to make sense of our exponential future. Every Sunday, I share my view on developments that I think you should know about in this newsletter.
Latest posts
If you’re not a subscriber, here’s what you missed recently:
🤳 Thanks to our sponsor Try It on AI, studio-quality AI portraits from the comfort of your home.
Sunday chart: Towards the positronic robot
In a leap that brings us closer to the robotic futures of sci-fi, Google DeepMind has unveiled RT-2, a vision-language-action (VLA) model trained on Internet-scale data that can be integrated into end-to-end robotic control. Previous robots have had to be programmed for specific tasks. DeepMind’s robotic agents learn how to do things from the Internet, generalise what they’ve learned to new tasks, accomplish these tasks, and even engage in rudimentary reasoning. Admittedly, they cannot do any of those things particularly well yet, but they are the first robots that can learn and “understand” (or at least, infer).
DeepMind has done it by combining robotic trajectory data with Internet-scale vision-language tasks like answering visual questions. The technical brilliance lies in transforming natural language responses and robotic actions into text tokens and incorporating them into the training set.
It’s significant because this is one of the the first time that large language models (LLMs) have been directly linked to the physical world with minimal human intervention. Previously, the way to make such a connection was through an API that linked the LLM to other applications that could then access the physical world. RT-2 creates the possibility of robots becoming more adaptable and independent, and the teaching of individual tasks a relic of the past.
These developments come as a refreshing break from the generally slow progress in robotics. Natural language-based AIs (especially LLMs) give robotics a new tool for progress that may jolt the whole field forward. As I argue in my book, new technologies intersect and enable fresh discoveries in other fields, especially if they are general purpose. Perhaps new AI regulation should incorporate Asimov’s three laws of robotics.
📸 Today’s edition is supported by Try It on AI, studio-quality AI portraits from the comfort of your home.
Last year a consumer portrait photoshoot cost an average of $750, while commercial shoots ranged from $800 to $5000, not to mention the weeks of planning, hair, makeup and styling. Today you can generate professional studio-quality images for just $17 in 1-2hrs from the comfort of your home. If you still haven’t experienced the power of Gen AI, join over 150,000 individuals and 500+ companies who are using Try it on AI to create professional AI portraits.
Exponential View readers get 15% off using EVTRY15 on checkout for any package. Code valid for the next 48 hours only.
Key reads
DIY superconductors. As we wrote last week, the buzz around LK-99 — whichever way the experiments turn out — has brought about the best of the collective intelligence of the Internet, as scientists from around the world conduct experiments with the limited information available in real-time and broadcast their findings on the Web. It’s a sort of mass peer review process mediated by public fora from Wikipedia to Reddit. You can tune into the prediction market which tracks public sentiment about the credibility of the breakthrough across the Internet. As of Friday, 4 August at 14:00 BST, it assigns a 40% chance of a replication before 2025, down from 48% on 1 August. The speed of experimentation around the world and the engagement with the story feels like a new era of science: we need scientific institutions and formal peer review, but DIY science and debate may become a much bigger part of how science gets done.
The stakes are high. A room-temperature superconductor could have huge implications for quantum computers, energy generation, and transportation.
Not on the same wavelength. Human-AI collaboration has been hailed as a silver bullet for enhancing efficiency, but an experimental study has revealed a surprising twist. Although AI alone was more accurate in diagnoses than two-thirds of radiologists, getting it to collaborate with human experts yielded no improvement. The authors call the phenomenon “automation bias/neglect,” and reckon it arises from doctors’ tendencies to underestimate AI’s insights. The takeaway is: giving human experts AI support doesn’t guarantee success. Successful collaboration means striking a balance between healthy scepticism of AI’s outputs (it may lack important context or hallucinate) and proper respect for its capabilities.
The Sino-American silicon curtain. The geopolitical landscape is shifting, and the increasing tensions between China and the US have placed Taiwan and South Korea (amongst others) in a predicament. As many readers will know, Taiwan is part of a “silicon triangle”, and Korea is strategically important because it’s home to important chipmaking and battery manufacturing companies. A shift occurred last year: South Korea exported more goods to the US than to China for the first time since 2004, a mark of changing alliances. Indeed, the US has been giving it subsidies in exchange for severing its links to China. Both Taiwan and Korea have advanced technologies that are instrumental to AI and climate, but not a lot of military might. As the US and China square off, they are being forced to pick sides. So far, signals seem to be pointing West.
Further reading: SpaceX is taking over the global satellite market fast, and this concentration of power is sounding alarm bells.
Data
Uber records a profit for the first time ever, earning $326 million.
The average American workday just got half an hour shorter, with productivity unchanged!
More than 75% of Internet bandwidth across the Atlantic, the Pacific, and Intra-Asia is used by content providers, a departure from the original intent of the Internet.
Humans can detect deepfake speech correctly 73% of the time.
Deforestation in the Amazon is at a 6-year low, dropping 66% YoY.
Short morsels to appear smart at dinner parties
🗝️ Cryptography-based tools could be a way to watermark AI-generated content.
🧪 Researchers are using AI to find extinct antibiotics.
🪖 A US military drone carried out combat tasks autonomously using a new AI.
🌍 IBM and NASA built an open-source Earth science AI model to track the effects of climate change.
🌆 How dense and built-up areas make heatwaves worse - the case for greener cities.
🐋 Could this ancient whale be the heaviest animal ever?
💟 Google docs is the new Tinder.
📚 Gone are the days of blunt book critics.
🧿 How to keep the coming mind-reading machines in check.
End note
I’m travelling for the next couple of weeks. While away, I have a few books going in parallel outside of my research reading — I turn to whichever I am in the mood for:
David Deutsch, The Beginning of Infinity
Richard Dawkins, The Blind Watchmaker
Ivan Turgenev, Fathers and Sons
Rodney Stark, The Rise of Christianity
Peter Attia, Outlive
What are you (re)reading this summer?
What you’re up to — community updates
Reid Hoffmann published a pushback on Wired’s generative AI policy, and another EV reader Gideon Lichfield offered a response.
John C. Havens co-wrote and led the IEEE paper “Prioritising People and Planet As the Metrics for Responsible AI”.
Ramsay Brown and his team at Mission Control published the second edition of Accelerate, a newsletter about quality, trust and velocity in AI.
Share your updates with EV readers by telling us what you’re up to here.
I just finished reading Outlive by Peter Attia. There is a lot of good detail on the longevity side of things - highly recommended if that’s your jam. But the main new take-away from me was in the final few chapters. Unexpected and high impact, not material I’ve seen other content around longevity look to address. Very glad I chose to read it. 👍
Read and Listened (3 times) Outlive recently, full of information and being an MD myself, had to go back and read and re-read certain chapters. Highly recommended for not only for reading but also frequent referencing.