Everyone’s been talking about America’s latest artificial intelligence technology these past few days — and the massive storm it’s supposedly about to unleash on markets. I think that’s complete nonsense. Every year there has to be a new concept to hype, creating a wave of money-raising fever in the markets.
US inflation came in worse than expected. The first half ended without meeting targets, and now the second half is a dead end — neither rate cuts nor rate hikes will do the job. The truth is, market expectations for Fed rate cuts have already been fully priced in. When cuts actually come, they’ll likely trigger significant market turbulence. So the Fed keeps stalling, hedging every statement. As I wrote last year: they won’t hand out money directly, but they’ll find indirect ways to stimulate consumption — because for stimulus to work, it has to reach the right people.
Which segment of Chinese society actually has spending power? People who own property. So cutting mortgage rates gives these people a little more room to spend. Mortgage debt itself, of course, is already a decades-long advance on residents’ future consumption.
Back to technological innovation: technology only generates real force when it forms an industry — not from isolated breakthroughs. The First Industrial Revolution happened in Britain. Why didn’t the Second happen in Britain too? Why did it happen in America instead? Think about it. The Second Industrial Revolution was built on railroads and telegraph systems. Americans raised massive capital from Europe to lay transatlantic telegraph cables and build a rail network stretching coast to coast. They had no money of their own — they fundraised heavily from Europe, especially Britain. There’s a book called The Crazy Investment that covers exactly this period. Over decades of constant investment, the protagonist went bankrupt multiple times, and along the way kept inventing new things to solve problems as they arose. Eventually, a telephone system was established on both sides of the Atlantic. Why did this happen in America, and later in Germany? Because they had the demand. Britain’s domestic market simply wasn’t large enough to generate that kind of need or that scale of opportunity. For any high-risk investment, the first question is always: is the market big enough? If the market is too small, it’s pointless — even success yields meager returns that may not even cover the sustained cost of investment.
Whether it’s the metaverse or whatever the current thing is, these are experimental concepts — their impact on social industry is far overstated. I actually think China’s recently announced breakthrough on the artificial sun project — 1,000 seconds of sustained fusion — is far more likely to drive a real technological revolution. Many people finish their PhDs and forget what they learned in middle school: conservation of energy.
Whether it’s neural networks or any other kind of network — every time you add one node, the number of connections across the entire network doubles. Connections consume energy, in case that wasn’t obvious. The human brain, even when doing absolutely nothing, consumes 40% of the body’s total energy.
A person who can’t stop overthinking — burning through energy internally — feels exhausted at the end of a day without having done anything. That’s because all that mental chatter is constant neuron-to-neuron signal transmission, a dense burn of energy. The smarter the system, the stronger its computing power, the more nodes it connects — which means the demands on energy and communications grow accordingly. The world’s most advanced communications infrastructure — in both theoretical standards and actual deployment — along with the intelligent chips that power it, is in China. Those companies are called Huawei, China Telecom, China Unicom, and State Grid.
From another angle: does artificial intelligence need data? Does it need consumer data or industrial data? Messy data or specialized data? The world’s largest industrial production base with the most complete supply chain — that’s us. The largest integrated e-commerce and social ecosystem — also us. Both useful and useless data lives here.
I’ve had this discussion with people before: what if the other side has such powerful algorithms that they can generate their own synthetic data to train their AI models? Generating data requires rules. Are those rules set by humans? According to human preferences? If so, what’s the point of training on data like that? You can abstract your own desires into a deity and worship it — but in that case, the deity’s state is entirely a reflection of the human’s state. If a divine force existed that humans couldn’t perceive, humans wouldn’t be interested in it and wouldn’t even know it exists. From a human perspective, such a god simply doesn’t exist. Understand this, and you understand why humans generating data according to their own preference-rules to auto-train AI application models is utterly pointless.
Try Google’s AI application and this becomes immediately obvious.