A.I. in non-English regimes and the things people say in…

China Wins Hands Down in EVs and Mega Projects

I say this out of deep respect for what it has achieved in the area of electric vehicles (EVs) and mega projects. Oh, the mega projects. I was looking at this timeline of how China created the first highway as early as the Qin dynasty, and then there was this whole period of civilisational peak and excellence around the Tang dynasty (one of my favourite periods in Chinese history, dare I say) and it became clear to me that the incredible work that China has achieved with such projects is next to impossible in non-collectivist cultures. It requires a tremendous willingness to place sufficient emphasis on benefits accruing to the group to undertake such public infrastructure projects. In spirit, it’s the same sort of collective action that’s required of lockdowns and combatting climate change. Which is why I genuinely feel like humanity will be on an unthinkable frontier in terms of progress if only, if only China can do for chips and compute what it did for EVs for the world. Just sayin’ and daydreaming, though.

Innovation Thrives in Open Societies

Of course, the boy can only cry wolf so many times and ain’t it good news for humanity in general that everywhere you look, innovations tend to thrive in free and open societies? Logically, it makes sense - you need the freedom of expression to expand the search space, and that openness is valuable for the sharing of learning, cross-pollination, spillovers, reducing and conditioning the search space on desired outcomes together. That they are using similar open source tools (and overwhelmingly ChatGPT rather than native LLMs) and that the solution sometimes lies in finding the right English prompts to use to me is further evidence of this and why closed societies will fall behind. In some respects, the more closed it becomes, the more it is at risk of repeating mistakes made in the Ming and Qing dynasties… I was hoping to learn of novel techniques like evolutionary mechanisms for model merging, or perhaps I was looking at the wrong places.

There are legitimate concerns and risks of A.I. for sure - but people’s knee-jerk reaction to go closed source lest open source software be used for harm by adversaries feels similar in quality to me to the secretive agent who will not share their billion dollar start-up idea with you lest you steal it. Capabilities matter, chokepoints exist in the exercise of A.I. (luckily) and infiltration is increasingly sophisticated and will only grow over time, yes; but sunlight is still the best disinfectant and defense against harm for the most part.

So I was most curious about A.I., and found myself walking away most intrigued by mapping the universe and extreme sports at this conference I attended in Hangzhou.

The Darndest Things That People Say Without Meaning to Offend

At some point during this series of talks about flying your own plane which I am interested in because I love travel, the speaker said something meme-able along the lines of, “I’m just an ordinary office worker. I flew a plane around the world and it costs me $110k USD back in 2016. You can do it too.”

This is such an interesting statement to make in a country where the monthly income of nearly half, or 600 million people in China is only 1000 yuan or ~138 USD.

I love it, and I suspect that there’s just not a lot of mixing in Chinese societies because even Mencius’ mum knew to move thrice to a good neighbourhood in the old days. I am always fascinated by the darndest things that Chinese people say and you know that they are not trying to offend, if anything our culture places an incredible emphasis on being a good host to any friend visiting from afar; my mum used to do unintended offensiveness a lot, perhaps even I do sometimes. As it turns out, not everyone would hire a driver to take them around for a day the way that I did - Tibet is popular as a travel destination amongst Chinese youths also because you can be extremely economical with it and camp outdoors to save on accommodation, without fear of getting chased away unlike in any other typical Chinese cities. So I am very well aware of the limits of empathy and trying to place myself in the shoe of another which is precisely why I think inclusion, representation and diversity matters - my suspicion is that such language is rare and hard to even broach in a place where 92% of the population is of Han ethnicity (that’s me) and that historically that’s always been an ongoing process of assimilation, it is made harder when the predominant narrative seems to equate harmony with sameness.

There’s so much to unpack when it comes to this unconference with Chinese characteristics, and at some point listening to all these talk of what skills are necessary in A.I. it just feels to me like the elephant in the room is really: what about political rights?

The Most Important Skill in the Age of A.I.

As in, let’s not even project any further growth in these technologies, let’s just take what we have and imagine a world in which what A.I. can already do permeates every single sector in the economy.

The future is already present, it’s just unevenly distributed. Politicians and corporate executives can spend all the time they want trying to understand what is this general purpose technology (while now knowing the fundamental difference between RAG, fine-tuning, model selection, evals, etc. and it is a very worthy question to ask indeed: how much can one regulate or make use of something that you just don’t know enough of? Or even better yet, what kind of structures or processes will we need to create to enable effective governance and deployment because while I don’t think we want to live in a world in which you have to be a PhD in A.I. to have a seat at the table to have a conversation about a technology that will affect us all, the fact remains that research and engineering inputs are critical to these decision-making pipelines) and debate what to do about it, but the adoption of A.I. is truly bottom-up. This is wonderful because its use is effectuated by actual need and use cases despite the amount of corporate handwringing or political gridlock that exists. It’s unstoppable. People are using it from the rooftop of the world in Tibet to paradise on earth in Hangzhou. People who know how to VPN their way uses ChatGPT because “it’s more powerful” while those who doesn’t use domestic LLMs. It’s unstoppable because it’s bottom-up, and if you chicken out of this race, you fall further behind, as simple as that.

If we play out this logic where someone can do 1000x the work through the effective use of A.I., it’s unclear to me why any company has an economical obligation to maintain a staff strength as before unless they are somehow pressured into doing so. I am not unoptimistic that in the long run we will find ways to adjust as a society and still have things for humans to do thanks to the magic of comparative advantage - even if A.I. beats us at every single task and is cheaper at doing them all, even if A.I. enjoys absolute advantage over a human worker, this podcast below explained the simple economics of why humans will still have things to do very clearly - but that adjustment, that interregnum is going to be extremely painful for some people. And there’s no telling whose throat the Damascus sword will cut.

All this is to say that we can debate soft skills, hard skills, or a mix all we want - but I think that the most important skill we will need as individuals that care about human lives flourishing in an age of A.I. is that of civic participation. So it is very fitting that I make this post on Labour Day.

Anyway, I will blog more about my trip to Tibet et. al. but for now, happy holiday world!

Originally published on PubPub at erniesg.pubpub.org/pub/7t92wlwf.