🔮 Asia pivots; controlling AI; humans vs bots, Bored Apes & aliens ++ #436
Your insider guide to AI and exponential technologies
Hi, I’m Azeem Azhar. 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, my team and I share our view on developments that we think you should know about.
Coming up
Before we commence with today’s newsletter, I want to share that on Wednesday, the 23rd of August, we’re releasing our latest Chartpack about the booming AI chip market, exclusively for our paying subscribers.
The need for specialised AI chips has never been higher than it is today. This fast-paced sector is projected to grow from $30 billion to a staggering $150 billion by 2027. Keep an eye out for our first overview of the market.
🎨 Thanks to our sponsor Masterworks, an investing platform that enables everyday people to invest in multimillion dollar paintings by artists like Banksy and Basquiat.
Sunday chart: Silicon’s Eastern frontier
Amidst the escalating U.S.-China geopolitical tensions and the rapid evolution in computing, the technological landscape of Asia is shifting. As traditional tech powerhouses grapple, India and Thailand are positioning themselves as contenders in the semiconductor sector.
In 2021, the Indian government greenlit a 760 billion rupee ($9.14 bn) program to bolster domestic semiconductor and display production, prompting U.S. chip giant Micron Technology to plan a new factory in the country. But there are still concerns about India’s infrastructure, such as power supply reliability. (We examined whether India could supplant China’s role in the technology supply chains a couple of months ago.)
On the other hand, Thailand is strategically pivoting towards front-end semiconductor processes, luring investors with tax incentives. While nations like Taiwan and South Korea traditionally dominated front-end chip processes, Southeast Asia was the go-to for the back-end. But given the U.S.-China strain, chip titans are now recalibrating their operations.
Meanwhile, China is grappling with its lowest levels of foreign investment since 1998. In an effort to reverse the trend, they introduced a 24-point plan to address concerns about protection of intellectual property and fair treatment in the government procurement process. Despite these efforts, China’s investment climate will remain clouded by unpredictable policies, geopolitical complications, and domestic economic concerns. In the U.S., the tech sector is split on President Biden’s restrictions on China, particularly in the chip domain. Leaders from Intel and Nvidia stress that Chinese market access is vital for the U.S. semiconductor industry.
🖼️ Today’s edition is supported by Masterworks.
A Banksy got everyday investors 32% returns?
Mm-hmm, sure. So, what’s the catch? We know it may sound too good to be true. But thousands of investors are already smiling all the way to the bank, thanks to the fine-art investing platform Masterworks.
These results aren’t cherry-picking. This is the whole bushel. Masterworks has built a track record of 15 exits, including net returns of +17.8%, +21.5%, and +35.0%, even while financial markets plummeted.
And the best part? Exponential View readers can skip the waitlist with the exclusive link below.
See important disclosures at masterworks.com/cd
Key (AI) reads
The geopolitics of AI. EV readers Ian Bremmer and Mustafa Suleyman highlight the challenges that AI presents to traditional governance. Due to its rapid progress and dual-use nature, there’s a need for a novel governance system grounded in principles like agility, inclusivity, and targeted approaches. They propose three governance bodies: a scientific risk-assessment group, a “Geopolitical Stability Board”, and a third body to bridge the gap between governments and the tech industry. I agree that we’ll need to manage geopolitical tensions, and have a body to establish a common understanding of emerging threats and how to manage them. This aligns with what I wrote about the need for global observatories to monitor, identify and tackle emerging threats. What matters, at this juncture in our technological revolution, is not achieving perfect consensus on AI governance, but rather establishing governance bodies that cultivate sufficient mutual understanding to promote cooperation and safety. This should be coupled with a diverse range of governance approaches and ongoing experimentation.
Are transformers a ladder to the moon? As we’ve written, our team uses LLMs extensively to help us in research and analysis (with a sharp eye on their limitations), and like many, we’ve been impressed with some of their behaviours that look like shallow reasoning. A recent paper suggests that transformers’ performance declines significantly with more complex reasoning tasks. This is because they tackle problems step-by-step rather than having a systematic understanding of the problem. Each step introduces the possibility of error, these mistakes compound in longer challenges, hindering the model’s accuracy.
OpenAI’s content moderation. OpenAI proposes using GPT-4 for content policy development and content moderation decisions. This would enable more consistent labelling, a faster feedback loop for policy refinement, and less involvement from human moderators. The promise is that you can easily apply and refine content moderation policies. Sound familiar? It’s a spin on Anthropic’s AI alignment approach, Constitutional AI. With this proposal, OpenAI argues that its AI can reduce the time and psychological demand on human moderators. Essentially, automating a tiresome and sometimes traumatising job. Yet, with big elections on the horizon, the question arises: should an AI have even more authority to decide who is allowed to say what? Algorithmic decision-making has a dark past.
See also:
A study shows that bots now outperform humans in “are you a robot” CAPTCHA tests. Distinguishing between real and artificial users is becoming increasingly challenging.
Defcon’s AI hacking event has helped find serious weaknesses in current LLMs.
The New York Times might sue OpenAI over an intellectual property dispute.
Weekly commentary
To understand the role of AI in our economic future, I’ve asked
to share his thoughts on how AI could power the Great Acceleration.🔮 Generative AI & economic growth
Hi, We have a special expert commentary for all members today by James Pethokoukis. I have been following James for a number of years - and was lucky enough to be on his podcast back in 2021. James is a senior fellow and the DeWitt Wallace Chair at the American Enterprise Institute (AEI), where he analyses U.S. economic policy. While I don’t always agree with the AEI, I’ve regularly found their analysis to be critical, challenging and additive to the debate. None more so than James’ work.
Data
Pirates ahoy! A basket of top U.S. streaming services will cost $87 this autumn, up from $73 last year, due to price hikes by major providers amid Wall Street pressure.
China has suspended the release of its youth unemployment data in July, following a record high of 21.3% in June.
Tesla supplier CATL unveils a battery that can charge up to 400 km of range in 10 minutes.
Generative AI has produced nearly 16bn images, three-quarters using systems based on Stable Diffusion.
Fewer tech unicorns are being minted than at any point in the past decade. Nearly half of them make physical products.
Short morsels to appear smart at dinner parties
🦠 Is this the year we eradicate Polio?
👽 Biology’s building blocks could help us find life on other planets.
🪄 Researchers theorise a hidden phase transition between solid and liquid.
🌘 Scientists have recreated Pink Floyd’s “Another Brick in the Wall” from neural signals.
👁️ Many are going back to Orwell’s 1984 to make sense of the world.
🐵 Buyers of Bored Ape NFTs are suing.
End note
I managed to finish three of the books I mentioned a couple of weeks ago.1 And I want to commend each of them:
Turgenev’s Fathers and Sons speaks of the tensions (a pre-exponential gap, if you will) between generations. Perhaps the struggle between tradition and progress is timeless. If you are reading the wondermissive (reader, you are), you’ll recognise many of the themes. A warning: Turgenev dragged me into the hole of 19th-century Russian literature. Currently on Dead Souls and Notes from Underground.
Dawkins’ The Blind Watchmaker is just a remarkable explanation of the mechanisms of evolution and how complexity arises from the process. I re-read it for the first time in more than three decades. This book should be on your must-read list.
Peter Attia’s Outlive gave me innumerable insights into developing one’s healthspan. I’m already putting several in practice. I hadn’t, for example, quite grokked just how critical VO2 Max was as a biomarker.
Azeem
P.S. 🔥🔥 Join me at the FT Weekend Festival on Saturday, September 2nd. EV readers get a special 25% ticket discount using the code EV23.
Sponsor us
To reach 260,000+ readers across our audience of executives, investors and founders, get in touch to explore partnership opportunities.
What you’re up to — community updates
Abishek Gupta wrote an article about the emergence of Shadow AI.
Gianni Giacomelli explains how generative AI can help you ideate.
Mike Zelkind’s 80 Acres Farms expand their distribution agreement with Krogers to over 1,000 stores.
Henry Coutinho-Mason launched VisuAIse Futures – an illustrated guide to AI innovation opportunities.
Share your updates with EV readers by telling us what you’re up to here.
To be fair to me, the reason I couldn’t get to the David Deutsch or Rodney Stark is that I forgot them at home in London.