🔮 Moon ladders; China’s tech policy; electric vehicles; blue hydrogen & Luddites ++ #335
Hi, I’m Azeem Azhar. I convene Exponential View to help us understand how our societies and political economy will change under the force of rapidly accelerating technologies.
Crypto has arrived as a political force. This is one of several implications of the skirmish between crypto lobbyists and the US regulators we've witnessed in the past couple of weeks; EV member, an erstwhile regulator with a deep understanding of blockchain, Kevin Werbach, discusses what is going on between the lines.
💌 I am offering a 50% discount (with extra goodies) on premium membership to Exponential View. At a special price of just $60 (instead of $120) for the first year of your annual subscription you will...
receive a gift copy of my book Exponential (£20 / $29 usual price),
receive my weekly Sunday wondermissive, the Wednesday overview of our world through data and other special essays during the month,
have a chance to participate in our Friday discussions,
be invited to join our closed Exponential Do community.
Enter promo code EXPONENTIAL50 during checkout.
*the offer expires on Thursday, Aug 19, and only applies to annual subscriptions.
Dept of the near future
Ladders to the Moon
🌗 Codex by OpenAI is a nifty AI programme that writes code for you. Enter what you want and voila you have an AI creating code for you. It is a wow moment... natural language to working code. It is not without limitations. Gary Marcus points out that the AI “is often correct on simple tasks, but frequently lost on more complex challenges.” (Jump back to my conversation with Gary where we discuss some of this.) Gary’s criticism notwithstanding, I can see tools like Codex helping many more people to become programmers. The shortage of talent is a major bottleneck for organisations trying to digitise. While you might not have a naive programmer using Codex build final code, the pairing of machine and human could get you a long way there – leaving more powerful programmers to polish the work off. (I am indebted to members of EV’s premium community for these insights.)
And as AI models get bigger - and better - they will lull us into a sense they know what they are doing until they don’t. It is the kind of problem we see with self-driving (as the recent Tesla crash reminded us, yet again).
Researchers have realised that large language models, like GPT-3 which underpins Codex, have plenty of benefits. (I talk about those models in a historical context here.) And so, groups are loading up on them. Consider that AI21 Studio’s Jurassic-1 Language Models is 100 times bigger than GPT-3, which was only launched 14 months ago.
That’s incredible growth in complexity. And it makes these tools more powerful and more useful. Industry has found that they are useful, and so is engaged in a process of exploiting the hell out of a design pattern that works. (And that is a pattern as old as the hills.)
But we still need to recognise that these are simply powerful tools, like spreadsheets or hammers or ladders. And this brings with it two issues. The first is that their complexity and power have significant bad effects as well as good. The first cars, limited to less than 15 miles an hour, did not need anti-lock brakes. Small AI models may not be problematic, larger ones could be. The second is the extent to which we mistake these AI systems (call them hammers or ladders) with something more. However big your ladder gets, it can’t get you to the moon. These software mechatrons might only get us so far. And focusing too much attention on them may distract us from genuinely novel approaches in developing AI.
See also: A South African court granted a patent to an AI system. My first response was to groan–only humans can invent, according to patent doctrine–but Meshandren Naidoo’s argument suggests the decision requires more nuance.
China’s tech policy takes shape
🇨🇳 The contours of China’s long-term tech strategy have been taking shape in plain sight for years. The recent crackdown on consumer-facing tech companies has, however, prompted a flood of articles about the future of technology in China. In EV333, I addressed some of the recent changes, arguing that China is redefining what progress in technology means for society.
In what might be the most complete piece to date on China’s technology policy, The Economist has gone into incredible detail arguing that the country will continue to evolve its own technology ecosystem with a strong focus on everything from artificial intelligence to self-driving cars. The consumer technology sector will exist but won’t feature as prominently as it does in the West. What is clear from the piece is that China is a country moving with intent and following a coherent tech strategy designed to firm up the country's power and prowess above all else. China’s path is divergent from that of the West but that shouldn’t overshadow what we can learn from how Beijing understands its role in the development of technology. [A great piece from Kai-Fu Lee on how China is using AI to fuel its next industrial revolution; I spoke with Kai-Fu about how China used the pandemic to bring its commercial AI and robotics closer to the isolated consumers].
The great transition
🚙 Automakers are finally investing in electric vehicles like they mean it. That’s the thrust of a new piece by EV reader Nathaniel Bullard in Bloomberg.
The transition from traditional automobiles to electric vehicles is easily one of the most fascinating industrial transitions in history. That’s because the car of the future has a fundamentally different architecture to existing combustion engine models. The future-auto is essentially a smartphone on wheels, uses a different powertrain, and increasingly relies on different business models (think subscriptions and services). The gravity of these changes might be one reason why over 2,000 Honda staff recently choose early retirement instead of retraining to work on electric vehicles.
Add to this mix the complex shifts in the supply line that has evolved over the last 50 years. Carmakers used to depend on OEMs, like Bosch and Valeo, part of the value-chain and ‘core competence’ thinking that pervaded the 1980s and 1990s. This means that they lost sight of their core suppliers. The shift to EVs is forcing them to rethink the visibility and anti-fragility of their supply chains. Small wonder that Tesla is making its own chips and batteries. [See also: The Economist on semiconductors, carmakers and the future supply chain]
Cities still matter
🌆 Talent agglomeration, especially around complex technical skills, is a topic I cover in my upcoming book. The phenomenon is particularly clear in cities around the world, where great minds come together and push innovation forward. That’s why we see tech companies clustered in places like Silicon Valley, New York, and Shanghai. This trend has been under pressure as more and more workers leave expensive cities to work from home during the pandemic and more governments try to spread the benefits of complex, advanced industries. And yet, as new research clearly demonstrates, cities continue to be the beacons of innovations.
🔋Dept of decarbonisation
CO2 level 414.60 ppm | 3,252 days until we reach the 450ppm threshold
The latest measurement of atmospheric CO2 (as of August 7, 2021): 414.60 ppm; August 2020: 413.56 ppm; 25 years ago: 360 ppm; 250 years ago, est: 250 ppm. Share this reminder with your community by forwarding this email or tweeting this.
🇺🇸 It’s a common trope to say a small portion of a large budget can fix a social ill for good. When it comes to the US defence budget, the numbers are staggering. Just 11% per cent of the Pentagon’s $716bn annual budget would be enough to produce wind and solar energy to power every home in America. Given the gravity of the climate crisis we are facing, these stats should take on a new meaning. The finances are there, it’s just about finding the right political capital to make sure they are used wisely.
⚡️Alternative energy sources like hydrogen are key to reaching global net-zero emissions targets. But how green are they? Remember when the Wall Street Journal recently found that the majority of solar panels used in America are made in coal-powered Chinese manufacturing plants? Well, in a new paper, researchers set out to determine how green is blue hydrogen. The answer is not at all. Blue hydrogen has worse carbon dioxide emissions than burning either natural gas or diesel for heat. ‘We see no way that blue hydrogen can be considered “green.”’ conclude the authors. Auke Hoekstra puts the power relations around H2 in perspective:
H2 is currently almost completely made from gas and the lobby is often paid for by gas companies that use the carrot of future blue H2 to greenwash all H2. This means there is no place for gas companies at the table.
Short morsels to appear smart while a robot makes your happy meal
💸 Are they real or fantasy? Why are DeFi yields so high (150% APY!)?
🕵️♀️ VCs are (slowly) getting more comfortable with disclosing their limited partners and the real money behind their funds.
🌱 Plant-based foods are maturing as the industry is projected to hit a $162bn valuation in the next decade.
💻 Fantastic essay on why CAPTCHA pictures are so unbearably depressing.
🛫 Amazon’s drone delivery dream is fading fast.
✈️ An essay on travelling, border crossings, and paperwork during Covid-19.
🦒 New research on giraffes reveals the complexity of their social interactions.
📖 Being called a Luddite might be considered an insult in some circles but the Luddite movement remains grossly misunderstood.
Back from my break, and I am starting to think about people doing interesting things in cellular agriculture, direct air-capture and solar radiation modification techniques. Happy to get some suggestions for who to speak to in the comments below. (Especially if you can connect me!)
Finally, join me on Aug 31 when I'll be speaking about the Exponential Age and how founders can stay ahead of the disruptive curve in a live event supported by Pitch, a new presentation platform. It's open to join!
What you’re up to – notes from EV readers
Lovely essay on the kindness and trust of strangers by JP Rangaswami.
Sophie Smallwood was quoted in “Melding flexibility and connectivity in the workplace” published by MIT Sloan Management Review.
Bill Dobie’s SEDNA closed a $34m Series B led by Rebecca Liu-Doyle at Insight partners to accelerate their mission to bring clarity and context to business communication within complex enterprises.
Great piece looking at Mike Zelkind and his company 80 Acres Farms. (Mike and I spoke about this at length earlier this year.)
To share your projects and updates, fill out your details here. Because of space constraints, we prioritise updates from paying members and startups I have invested in. (You can become the former by subscribing, if you have not already, and the latter by getting an intro to me via a trusted contact.)
Featured jobs for readers of Exponential View
Members of Exponential View are hiring!
Sustain.Life, a high-energy software startup that specializes in B2B solutions for corporate and SMB sustainability efforts, is looking for a Product Design Lead.
Stripe Climate is looking for a Full-Stack Engineer.
Planet Labs, an aerospace and data analytics company, is looking for a Director of Communications.
Switchfly, a travel-commerce and loyalty SaaS platform, is looking for a Senior Software Architect.
If you're hiring, let email@example.com know.
How happy are you with today's edition?