Hacker News Reader

Comment by TekMol | Hacker News Reader

With AI, will China still be the center or production? Or will production take place close to where the produced goods are needed?

As I understand it, most things are produced in China because of low wages. But AI has the same wage everywhere around the world: 0.

View on HN

That’s really not so true any more. The number of industrial robots is higher in places like Hong Kong than the US (and China as a whole is not far behind in spite of their massive population) while places like Singapore and South Korea has massive robotics per capita. China has decent wages in many places, higher than much of the world, and they have pretty strong agglomeration effects now, improving manufacturing efficiency.

Not to take away from the Asian countries, but this is a story about Germans putting together their own panels. It’s cool to see some variety springing up. One can support their own manufacturing tech and save on transportation.

As I understand it, it's not just low wages but also the supply chain ecosystem. Anything you could possibly need for high-tech product manufacturing is available locally in Shenzen.

Considering that the pandemic has effectively broken overseas supply chains, this would be the perfect time to switch to domestic production. It would help the job markets too.

Most international supply chains are unaffected by the pandemic. Looking at the UK in particular, Brexit had a bigger impact on supply chains than the pandemic, but even there (IIRC the UK is short of about 100k truck drivers and having problems even with some domestic supply chains) the international supply chains are mostly still in place.

We are going to see major shortages in the coming months. I work in IT infrastructure and we were told by suppliers that there are going to be major delays in equipment orders due to the pandemic disrupting the supply chain. This is going to affect multiple industries for the next year or so.

Sure, it's zero if you ignore all the development and maintenance costs.

Plus, robots aren't necessarily better at making things. Tesla, for example, rolled back its fully automated production line because it wasn't as adaptable as humans are.

Wages aren’t the only cost, and it isn’t clear how good and AI would need to be to fully replace production line workers — it might be that an AI good enough to do that would need to also be good enough to render all human labour obsolete, including ours.

Many companies will not have AI systems, the few that do will rent it just as they do with their other software. It is not free, no, and won't be.

> But AI has the same wage everywhere around the world

not really - look at BTC.

Can you explain what you mean? As I understand it, BTC is tied solely to USD for exchange.

i mean bitcoin mining locations is energy price dependent, and the same applies for AI which is very energy intensive too, and thus AI "wage" is going to be different across the world.

Ah that's a very good point. Thank you!

so - places where you can get away stealing electricity are going to have an edge.

kinda like china with bitcoin mining...

> same applies for AI which is very energy intensive too

Training (building the model/AI) might be, inference (using the model to actually "run" the AI) is relatively cheap, I think. This may change with massive models, but I suspect power usage of the AI will not be a major concern for deciding where to put AI-operated factories.

training once vs. millions of inferences. And tomorrow inference may naturally include incremental training. Human brain consumes 20% of the body energy. I don't think AI will come to human brain energy efficiency soon, and thus i think AI (and its accessories, ie. like 3d/AR/VR model spaces where and what AI will be operating with) will be a major, if not main, energy consumer going forward. When Google self-driving car passes by the main sound coming from it isn't the engine nor the tires :)

Reply on news.ycombinator.com