Don't Outsource Your Understanding

Last June, Ford’s CEO Jim Farley said in an interview with Walter Isaacson, “AI will leave a lot of white-collar people behind.” He then said at the Aspen Ideas Festival that half of white-collar workers could lose their jobs to AI in coming years. By October 2025, Ford’s COO Kumar Galhotra said Ford was “deploying AI across the entire industrial system”. They inserted 900 AI-powered cameras to “detect quality issues at the source and help us mitigate supply disruptions”. It then fired their quality inspectors.

Ford now says it is rehiring 300-350 “veteran” quality inspectors although they have re-titled them “engineers”. Charles Poon, VP of vehicle hardware said, “Artificial intelligence is a fantastic tool, but it’s only as good as the information you use to train it. Mistakenly, we thought that by just introducing artificial intelligence and ingesting the design requirements we had, that would produce a high-quality product.” Veteran quality inspectors had built up knowledge over many product cycles that were not in the design documentation and as a result quality went down. They are now using their veteran workers to train up their AI systems and mentor younger workers.

Another anecdote is from Arnal Dayaratna of IDC, who noted LLMs are terrible at writing. “Transitions between paragraphs tend to be terrible. As a result, LLMs tend to generate prose that is absolutely unreadable, and this is especially true if you are writing about a topic that is new or cutting-edge in some way, as they have very little in the way of a formal training dataset to draw upon.”

The issue is that AI is not “intelligent”. You can scaffold around AI and make it seem intelligent. There are techniques like RAG (retrieval augmented generation) that can ground AI and make it do wondrous things. But ultimately a human put all that scaffolding together. Without the intelligent design of the human, the AI would not be so useful. The idea that AI will create end to end solutions and replace humans on its own seems like a marketing spin designed to get high valuations.

But is the absence of the advertised super-intelligence, which will create a utopia where no one needs to work, a bad thing?

To try to answer this question let’s travel to China, where Chinese scholars have been applying a Marxist lens to questions like “Who or what creates value after the AI revolution? The machine? The human who invented it? The human who operates it?’

In April, a Hangzhou court answered the philosophers and ruled against a Chinese tech company which had replaced a quality assurance worker with AI software. The court said, “The development of AI technology should be applied to liberating labour, promoting employment and improving people’s livelihood. Labor law allows employers to undertake technological changes and upgrade their operations, but it should also take into account the protection of workers’ legitimate rights and interests”. The Chinese government has also warned companies against replacing workers with AI.

A 2024 Ipsos Global Survey said 83% of Chinese people believe AI-powered products and services offer more benefits than drawbacks, compared to 39% in America and 55% globally.

In medicine, a Shanghai-based doctor said an AI assistant organizes patient information and flags warning signs before the consultation even begins, but "it doesn't replace clinical judgment... it can make the system much more efficient." China's National Health Commission formalized this stance in a November 2025 guideline calling for AI to support the full continuum of prevention, diagnosis, and rehabilitation.

In acupuncture, AI-powered pulse-diagnosis machines and tongue scanners are translating centuries-old tactile diagnostic techniques into quantifiable data. Algorithms suggest acupuncture points based on symptom patterns. But the literature is explicit that this is meant to enhance practitioners' pattern-recognition, not substitute for their judgment. It stresses "human touch", including intuition, empathy, physical dexterity, remains indispensable to the practice.

Shenzhen's court AI system (covering 85 judicial procedures) helped judges handle 50% more cases in 2025, but the court was explicit that judges remain responsible for the final decision at every step. In one case, a judge had AI draft over 70% of a judgment but still needed to refine language and reasoning by hand.

China Agricultural University's Shennong AI Agent Platform helps farmers identify crop disease (e.g., corn leaf spot in Heilongjiang) and get treatment recommendations in real time, while a parallel Wheat Breeding AI Assistant supports researchers rather than replacing their expertise. Official commentary explicitly notes automation of repetitive or hazardous tasks lets older farmers manage larger fields with less physical strain, while new higher-skilled, better-paid roles (drone operators, data analysts) are created around the technology.

Ironically, because AI has a high failure node, as Ford witnessed, U.S. enterprises are now learning the hard way that augmentation is superior to replacement and moving towards the Chinese approach.

This helps the Chinese ecosystem, and more importantly creates the potential for domination in AI models. Ben Cera, CEO of Polsia AI, a U.S. AI start-up company, argued that China’s open-source models perform nearly as well as leading frontier models at a fraction of the cost making it hard for companies to spend $100 million when similar capabilities could be available for $5 million. “China will crash the U.S. stock market this year. The bubble isn’t AI. It’s the price of AI. And it’s about to pop”. While Silicon Valley dismisses this narrative, it’s hard to see U.S. enterprises being generous with AI spend if they can’t fire scores of workers.

Blue Owl Investors Feel Blue. Investors in two Blue Owl Capital flagship private credit funds requested $4.7 billion of withdrawals. One fund, Blue Owl Credit Income Corp (OCIC), saw redemption requests equal to 19% of its shares outstanding. The technology-focused fund, Blue Owl Technology Income Corp (OTIC), saw requests for 38%. OTIC’s redemption rate is roughly 2x-4x higher than peer BDCs such as Apollo, Blackstone and Goldman. Redemptions have been high on heightened market concerns around AI-related disruption to software companies. While the absolute numbers remain large, redemption requests slowed down quarter on quarter suggesting, perhaps, that the worst of the redemption phase around software is over. That could be a positive signal for software equities.

Shopping Around. The Official Monetary and Financial Institutions Forum published a report saying central banks view the euro and yuan as attractive alternatives to the dollar “with advanced economies favouring the former and emerging markets focusing on the latter.” In ten years, central banks expect to hold 52% in dollar assets, 23% in euros and 5% in yuan. For the euro, 55% of those polled by the OMFIF say “permanent, large-scale EU” debt issuance rather than individual country issuance would raise their willingness to hold euro reserve assets.

The yuan percentage seems small but given that China does not have an open capital account, it's not a surprise. Central banks hold reserves based on depth, liquidity and convertibility of the underlying asset markets. China has an offshore yuan market, but the 'dim sum' bonds only represent $179 billion of issuance versus $11.9 trillion in China's onshore bond market. China has been accumulating gold and trying to deepen its gold exchange infrastructure. This could extend a mechanism that already exists in nascent form where African sovereigns like Egypt have issued yuan-denominated 'panda bonds' because they lack sufficient access to dollar markets on reasonable terms, currently relying on multilateral guarantees from the African Development Bank and AIIB. A gold-backed claim would be a novel substitute for that guarantor role, letting China itself absorb the convertibility risk that currently keeps non-Chinese sovereigns wary of yuan debt. This in turn could give central banks a genuinely liquid, gold-backed yuan asset to hold as reserves, addressing the depth and liquidity gap that currently cap yuan's reserve share. I’m not saying this will happen. It will require China to have far more gold reserves and take years to establish, but it’s something to watch for in the news in the years ahead.

The American Revolution. Melat Kiros, 29, a democratic socialist and first-time candidate, toppled 15-term Representative Diana DeGette in Denver’s 1st District. In New York, three progressive candidates endorsed by Zohran Mamdani won House primaries. Brad Lander defeated two-term Representative Dan Goldman in the 10th district. Darializa Avila Chevalier narrowly defeated five-term Representative Adriano Espaillat. Espaillat was chair of the Congressional Hispanic Caucus. These are deep-blue districts. Moderates prevailed in districts that are more competitive such as Micah Lasher prevailing over progressive Alex Bores in the 12th district.

Three upcoming races will determine whether the democratic socialists become an intra-party faction or open a real lane for someone like AOC in the 2028 presidential primary. In Michigan, progressive Abdul El-Sayed is making a run in the Senate primary scheduled for August 4th. Graham Platner is a progressive running against Republican Senator Susan Collins in Maine. In past Reddit posts the oyster farmer sounded like a democratic socialist, but he’s now asserting he is more a New Deal Democrat. And progressive Francesca Hong is running for Wisconsin governor. If these races swing to centrist candidates, the most likely 2028 Democratic presidential nominee should be someone more centrist with a mandate to build a more moderate governing coalition.

End Note

Monaco experienced its first-ever mafia-style bombing in a residential building. The perpetrator attacked a Ukrainian under sanctions who is known for selling liquor in Crimea. It’s unclear why he was targeted and the attacker fled the principality. While this is the first (and hopefully last) attack, the film industry has long featured Monaco as a place for hits.

In Iron Man 2, Ivan Vanko, known as Whiplash, attacked Tony Stark on the Grand Prix circuit with electrified whips, turning the glamorous race into a battlefield. In GoldenEye, a terrorist hijacks a military helicopter in Monaco, and in Never Say Never Again, SPECTRE sets their operations in Monaco.

Monaco, however, is a very safe country in real life. There are no terrorist attacks, assassination attempts, spy operations, heists or car and motorcycle chases. You are far more likely to see a lady with her poodle. Hopefully that stays the same for the immediate and long-term future.

For all my U.S. readers, have a Happy July 4th!

Omar Sayed