A Few Things: Leaked Google Memo on Future of AI, Modi's India, Dan Wang on China, Frankopan's Latest, The Impact of Gen AI, The Future of Software, Your Personal AI, News & Charts You Missed...
May 19 2023
I am sharing this weekly email with you because I count you in the group of people I learn from and enjoy being around.
You can check out last week’s edition here: The Sohn Conference, What Climate Alarmists Are Getting Wrong, Spinoza, What's Next In The Chip War, A New Game Plan For Venture, News And Charts You Might Have Missed....
Believe it or not, that “♡ Like” button is a big deal – it serves as a proxy to new visitors of this publication’s value. If you enjoy this article, don’t be shy :)
Quotes I Am Thinking About:
“Almost every man wastes part of his life in attempts to display qualities which he does not possess, and to gain applause which he cannot keep.”
- Dr. Samuel Johnson
“You can tell more about a person by what he says about others than you can by what others say about him.”
- Audrey Hepburn
"Learn to listen. Opportunity sometimes knocks very softly."
- H. Jackson Brown Jr.
“True happiness is to enjoy the present, without anxious dependence upon the future, not to amuse ourselves with either hopes or fears but to rest satisfied with what we have, which is sufficient, for he that is so wants nothing. The greatest blessings of mankind are within us and within our reach.”
“My favorite 11th commandment is, ‘Thou shalt not take oneself seriously.’ You need to constantly challenge your own thoughts.”
- Sam Zell. RIP
A. A Few Things Worth Checking Out:
1. The Economist had a great article titled: What does a leaked Google memo reveal about the future of AI?
A Google memo notes….that researchers in the open-source community, using free, online resources, are now achieving results comparable to the biggest proprietary models. It turns out that LLMS can be “fine-tuned” using a technique called low-rank adaptation, or LoRa. This allows an existing llm to be optimised for a particular task far more quickly and cheaply than training an LLM from scratch.This could have seismic implications for the industry’s future. “The barrier to entry for training and experimentation has dropped from the total output of a major research organisation to one person, an evening, and a beefy laptop,” the Google memo claims. An LLM can now be fine-tuned for $100 in a few hours. With its fast-moving, collaborative and low-cost model, “open-source has some significant advantages that we cannot replicate.” Hence the memo’s title: this may mean Google has no defensive “moat” against open-source competitors. Nor, for that matter, does Openai.
This has both positive and negative implications. On the plus side, it makes monopolistic control of ai by a handful of companies far less likely. It will make access to ai much cheaper, accelerate innovation across the field and make it easier for researchers to analyse the behaviour of ai systems (their access to proprietary models was limited), boosting transparency and safety. But easier access to ai also means bad actors will be able to fine-tune systems for nefarious purposes, such as generating disinformation. It means Western attempts to prevent hostile regimes from gaining access to powerful ai technology will fail. And it makes ai harder to regulate, because the genie is out of the bottle.
2. Controversial London Review of Books article by Pankaj Mishra titled: The Big Con: Modi’s India and the New World Order.
Key bits (emphasis mine):
Modi has counted on sympathetic journalists and financial speculators in the West to cast a seductive veil over his version of political economy, environmental activism and history. ‘I’d bet on Modi to transform India, all of it, including the newly integrated Kashmir region,’ Roger Cohen of the New York Times wrote in 2019 after Modi annulled the special constitutional status of India’s only Muslim-majority state and imposed a months-long curfew. McKinsey’s global managing partner, Bob Sternfels, recently said that we may be living in ‘India’s century’. Praising Modi for ‘implementing policies that have modernised India and supported its growth’, the economist and consultant Nouriel Roubini described the country as a ‘vibrant democracy’. But it is becoming harder to evade the reality that, despoiled by a venal, inept and tyrannical regime, ‘India is broken’ – the title of a disturbing new book by the economic historian Ashoka Mody.
The number of Indians who go to sleep hungry rose from 190,000,000 in 2018 to 350,000,000 in 2022, and malnutrition and malnourishment killed more than two-thirds of the children who died under the age of five last year. Meanwhile, Modi’s cronies have flourished. The Economist estimates that the share of wealth held by billionaires in India that derives from cronyism has risen from 29 per cent to 43 per cent in six years. According to a recent Oxfam report, India’s richest 1 per cent owned more than 40.5 per cent of its total wealth in 2021 – such statistics are more often associated with the notorious oligarchies of Russia and Latin America. The new Indian plutocracy owes its swift ascent to Modi, who audaciously clarified the quid pro quo. Under the ‘electoral bond’ scheme he introduced in 2017, any business or special interest group can give unlimited sums of money to his party and keep the transaction hidden from public scrutiny.
Like the Russian elite, Modi and Adani have succeeded in bending the moral arc of politics and journalism towards greed. Jo Johnson, who had to disentangle himself with haste from Adani’s global cash nexus, was, as a reporter for the Financial Times, a rare practitioner of sober Western journalism on India, at a period when opinion-making periodicals such as Time, Foreign Affairs, Newsweek and the Economist were hailing the country as a ‘roaring capitalist success story’. ‘Unless India makes a dramatic investment in its human capital,’ Johnson wrote in 2006, ‘its demographic advantages will turn into a demographic disaster in the form of a massive unemployable labour force.’ His prognosis has become even more menacing today as the country’s population overtakes China’s, the scope for labour-intensive jobs in Indian industry shrinks further, the large middle class long fantasised about by foreign corporations stubbornly fails to materialise, and private investment keeps falling despite lavish government spending on infrastructure. Modi’s government has not spent the sums on public health and education that, as Johnson observed, would be necessary for securing a demographic advantage. Instead, it has sought to deploy many of the ‘unemployable labour force’ as stormtroopers of Hindu supremacism, indoctrinating them in a garishly fabricated version of the Indian past and equally kitsch daydreams of its future as a world guru.
3. If you are curious about China, you have most likely heard of read Dan Wang’s annual letters, he covers technology for Gavekal Dragonomics.
He spoke with Noah Smith about China's economy, decoupling, export controls, industrial policy, state control, and lots more.
Here are key bits:
N.S.: That's a shame. But in exchange, I'm hoping you'll give us some more detailed analysis of Chinese industrial policy and technological competitiveness, of the type you used to only provide in your private research!
So my first question is: What are two or three things that Americans need to know about Chinese industrial policy and technological competitiveness in 2023?
D.W.: The first is that there have been a lot of failures. China is achingly aware of its deficiencies in two strategic sectors in particular: semiconductors and aviation. So it has showered these sectors with bountiful money and stern policy attention. Where has that gotten them? Not far. On chips, China has built the basics of the industry, but is at best 10 years behind the leading edge of manufacturing logic chips, and even more on the tools needed to produce chips: lithography equipment and EDA software. On aviation, China’s answer to Airbus and Boeing has been years behind schedule, and is anyway substantially dependent on western engines and avionics systems.
The second is that there have been a lot of successes. I would say that China has caught up with the west on nearly all manufactured products outside of chips and aviation. It is making sophisticated electronics components. It is making boring industrial equipment that rarely grace headlines. And it is making most of the technologies we need for decarbonization. The folks at Bloomberg New Energy Finance estimate that China owns 90% of the solar supply chain: everything from polysilicon production to the tools needed to make photovoltaics to the panels themselves. It’s also doing very well in batteries and has a shot at dominating the hydrogen supply chain as well.
A central question I’m working on now is to figure out whether China’s technological capabilities in the future will more closely resemble its failures or successes. There are enormous macro headwinds for China’s economy: lousy demographics, a domineering central government, and greater political emphasis on state-led growth. But I tend to think that China will be able to solve its technological deficiencies. Its main task is to reinvent existing technologies: arguably firms like TSMC and ASML have the harder task, that is to push forward the technological frontier. On chips in particular, there’s a broad sense that it costs too much to keep pushing forward Moore’s Law. So if the leaders hit a wall, it’s only a matter of time for Chinese firms to catch up. These products are technology, not magic. And Chinese firms have already mastered a lot.
N.S.: What's different about semiconductors and advanced aviation? Why has China not been as successful in catching up in those sectors?
D.W.: I think there has been a consistent pattern of Chinese successes and failures. Any technology that demands the complex integration of different scientific areas is challenging for Chinese firms. Semiconductors bring together electrical engineering, chemistry, computer science, and more; aviation is the integration of aerodynamics, materials science, mechanical engineering, etc. China's scientific capabilities have steadily risen, but I would say it's still fairly weak. No surprise, perhaps, that Chinese firms weren't able to produce mRNA vaccines, since its scientific establishment is unused to puttering around the fringes of new fields.
On the other hand, for any technology where the science is mature, and the complexity lies more with the manufacturing process, China tends to be strong. Take renewable technologies like solar photovoltaics or EV batteries. The science of turning light into electricity and power storage are pretty well understood. But Chinese firms have been able to outbuild their foreign competition (with plenty help from government support) in creating high-performing products. Putting together a battery, for example, involves around ten steps—from cell filling to final sealing—that demand perfect handoff at each stage. Chinese firms are really good at this, which they learned from the highly-demanding electronics supply chain.
And here the US tends to be weak. American manufacturers aren't good at making products of high intricacy at high volume. And it sometimes trips over simple products too. It’s puzzling to me that American factories weren't able to quickly retool to turn out masks and other personal protective equipment in the early days of 2020. There's something quite strange about the US where it is able to make super-advanced products like AI, jet engines, semiconductor production equipment, but can't build basic infrastructure or simple products.
N.S.: One interesting thing that China's leaders have tried to do is to actively discourage industries they want to see shrink. A lot of people were astonished at the government's crackdown on the IT industry in 2021. Now a lot of people are saying it's been substantively reversed. Would you agree with that assessment? And if so, will there be lasting scars from that incident? Can we expect other crackdowns on specific industries in the years to come?
D.W.: The state has relented. But it hasn't reversed. As my colleagues at Dragonomics have written, Beijing's regulations on platform companies are now entrenched at a permanently burdensome level. The good news for these companies is that future regulations are more likely to be marginal. The bad news is that the golden age of internet platforms is over: they won't again have such an easy time making enormous profits in a lightly-regulated market.
These regulations have indeed left scars. Entrepreneurs in sectors that include ride-hailing, video games, and especially online tutoring were left shell-shocked. But pain set in for sectors that went beyond those directly targeted by the state. A common sentiment from an entrepreneur goes along the lines of: "I would like to focus on building a business and working on my product; instead I'm being forced to become a political scientist to understand the direction of policy."
The major question today is whether these scars will persist or fade over the next few years. It's undeniable that a lot of entrepreneurs were angry about the crackdown; and there are many anecdotal reports of wealthy Chinese who have decided to move to Singapore. Beijing has shifted its tone since the end of 2022, saying the most soothing words about how much it loves private businesses. But who can blame entrepreneurs for feeling skeptical, after many of their creations were strangled by the state? On the other hand, if China's growth is impressive this year and next, I expect that many of them will return. It's not obvious that they can establish themselves overseas, after all: the US has not gone out of its way to rhetorically welcome Chinese entrepreneurs. If growth can pick up, and Beijing can pause its political tightening, then scars will fade more quickly. But these are uncertain ifs.
B. The Earth Transformed
Peter Frankopan is a British historian and writer. He is a professor of global history at Worcester College, Oxford and the Director of the Oxford Centre for Byzantine Research. He is a fellow of the Royal Asiatic Society and is best known for his 2015 book The Silk Roads.
In The Earth Transformed endeavours to tell the story of homo sapiens from the perspective of the world we inhabit, the disasters we survive and the impact we create.
The FT said: Frankopan distils a mass of historical sources and scientific data spanning thousands of years into an epic and spellbinding story
History as usually told centres around leaders and dates. Peter Frankopan’s thesis is that these are just pawns on a much larger, ever changing chess board. That board being Earth and its changing environment. The book isn’t about climate change but how easily our species is impacted by changes in its environment and how often these events occur.
On page 10, Frankopan states:
In the main, though, we ignore climate and long-run climate patterns or changes altogether when we look at history. Most people can name great leaders and major battles in the past, but few can name the biggest storms, the most significant floods, the worst winters, the most severe droughts, or the ways that these influenced harvest failures, provoked political pressures or were catalysts in the spread of disease. Reintegrating human and natural history is not just a worth while exercise; it is fundamentally important if we are to understand the world around is properly.
Frankopan has three goals for this book:
Re-insert climate back into the story of the past as an underlying, crucial and overlooked theme in global history.
Set out the story of human interaction with the natural world over millennia and loo at how our species exploited, moulded and bent the environment to its will both for good and for ill.
Expand the horizons of how we look at history beyond just the periods humans have occupied.
A great quote from the book:
However, periods spanning decades, centuries and even millennia are the briefest of blinks of an eye in terms of history of this planet; and in that sense they are perhaps not even important ones either; the earth has been spinning around the sun for billions of years, and will continue to do so for billions of years more. In one way, then, the Holocene is the perfect tag for the modern age - a time that is geologically and climatically distinct, but which also corresponds to the era of modern humans, an era we invariably fail to place within he bigger picture.
The book does a good job bringing together historical changes in our environment to the need for migration, trade and belief in divine power, connecting many historical events that define our world today.
I often look at various forecasts for 2050 or 2100 and think it’s just an example of crystal ball gazing and catastrophising, but Frankopan argues that it’s less hard to argue with problems that are already here. This is not about something that will occur tomorrow, but something happening today.
For example, taking something as simple as air pollution, in some cities it is at levels 10 times higher than WHO minimum standards, and 92% of the world’s population live in places which exceed these limits. Air pollution is surprisingly lethal, in 2015 it resulted in 9 million premature deaths worldwide.
Today, we have the highest global carbon dioxide concentrations for the past 2 million years. The rate of change is particularly significant. In 1750 (the beginning of the Industrial Revolution) there were 250 parts of CO2 per million air molecules; now it is 425, causing change in all our ecosystems.
And CO2 isn’t our only problem. Take food waste for a second. In the UK, almost 10 million metric tonnes of food (£20bn worth) is thrown away each year. Getting this food from the field, pasture, coop to the rubbish bins creates 36 million metric tonnes of greenhouse gas emissions.
But human history is a Story of Survival. Mankind has roamed the planet for only 0.00001% of the earth's existence.
As descendants of life forms that survived five mass extinctions, the human story is one of iterative failure, survival and adaptation. Great empires (eg. Nineveh, the Roman Empire) had periods of glory before failing, just as many companies have (ie. RCA and Kodak).
Given all this survival, why worry about climate change? Frankopan sees climate change as the ‘straw that breaks the camel’s back’, more than likely catalysing a catastrophe of some sort.
While numerous things (asteroids, volcanoes, pandemics, nuclear war) could cause a major population collapse in theory, in practice such events ‘tend not to happen’. That said, populations are declining everywhere in the developed world.
But regardless of a huge catastrophe, one of the unfair things about climate change as it stands today is that it’s poorest the suffer the most with climate change, which dramatically increases the risk of civil unrest and radicalisation. The effects are not merely climatological and economic but also political.
But could technology fix all our problems, after all humans have been innovation and adapting for centuries?
Here he states:
It is not impossible, of course that humanity proves able to adapt, perhaps by changing the we live and the choices we make; perhaps thanks to new technologies and ideas; perhaps through higher levels of collaboration, brought on either by enlightened governance or by crisis and necessity. It is worth bearing in mind, however, that much of human history has been about failure to understand or adapt to changing circumstances in the physical and natural world around us, and about the consequences that ensue.
If you want to get the main ideas from Frankopan rather than reading the +600 page book, watch this recent Channel 4 interview or hear him discuss it on the Rest is History podcast.
Thank you Mark Evans and David Giampaolo for the encouragement to power through the book and hear Peter discuss his book in person.
C. The Technology Section
1. What will be the impact of AI on our lives? That is probably too broad a question to be viable.
Question 1: What’s the impact of AI on the economy and jobs:
Generative AI could power a productivity boom. According to Erik Brynjolfsson and colleagues at the Stanford Digital Economy Lab we could have a AI-powered productivity boom due to an increase in efficiency and an acceleration of innovation.
Generative AI has broad applications that will impact a wide range of workers, occupations, and activities. Unlike most advances in automation in the past, it is a machine of the mind affecting cognitive work.
This super piece titled: Meet Your New Finance Intern: GPT-4 goes into detail of what it would look like.
Question 2: Double click further, What will be AI’s impact on software which is so central to all our lives?
Disclaimer: I am a small LP in their fund.
2. The world seen by Gen Z’s is very different from other generations. This Rex Woodbury piece does a great job discussing AI Chatbots and Our Loneliness Epidemic.
3. One of my strong views is that AI’s Next Big Thing is Digital Assistants. This Daniel Miessler goes into details and this Elad Gill podcast with With Mustafa Suleyman) Founder of DeepMind and Inflection) asks Will Everyone Have a Personal AI?
4. The popular GLP-1 class of weight loss and diabetes drugs was also shown to boost the activity of cancer-fighting immune cells known as natural killer (NK) cells in a small clinical study. The increase in function of NK cells was independent of weight loss, suggesting an independent mechanism.
D. News and Charts You Might Have Missed:
1. Young Americans are dying at alarming rates, reversing years of progress. The US is the only place among peer nations where firearms are the #1 cause of death in young people. Transportation-related fatalities have also risen significantly, owing partly to distractions from cellphones, reports say.
2. Taylor Swift’s Eras Tour is expected to be one of the highest-grossing tours of all time, but its impact reaches beyond tickets and merchandise, as fans spend on travel, lodging, clothes and makeup to attend.
Her fans’ spending could point to a larger trend of consumers indulging in experiences they missed during years of cooped-up pandemic living. Each stop on the Eras Tour is preparing for a boom as fans cross state lines to attend whatever shows they could get tickets for.
Glendale, Arizona, where hotel rates soared when fans visited from around the world, temporarily renamed itself “Swift City.” Swift’s 3-night stay in Houston resulted in the city’s highest hotel revenue week of 2023, jacking up prices more than the NCAA’s Final Four tournament. The Courtyard by Marriott in Philadelphia, near the 65k-seat stadium where Swift performed recently, is completely booked (a rare occurrence). Fans want to look good for the shows, so they’re shelling out for bejeweled boots, custom jackets, and hair and makeup. And major retailers like Amazon and Poshmark are selling out of Eras outfits.
3. The world will add 1 billion air conditioners before 2030, many in places in the developing world, and many in India in particular. Economists observe that a country tends to add air conditioners en masse when household incomes cross $10k, and when adjusted for purchasing power India’s at $9k this year. At Daikin Industries, the largest AC manufacturer in the world, sales are up 15x what they were. A study of thousands of Indian factories found productivity dropped 2% for every degree of Celsius increase in the temperature of the factories.
4. Get me out of here….
5. The latest LLM’s like Claude from Anthropic can take entire books as inputs to analyse.
Believe it or not, that “♡ Like” button is a big deal – it serves as a proxy to new visitors of this publication’s value. If you enjoyed this, don’t be shy.