This post was originally published on this site.
Watching the global frenzy about the new Chinese artificial intelligence model, DeepSeek, last week, I remembered the Statement on AI Risk published by the Center on AI Safety and signed by 350 experts, on May 30, 2023.
It’s an archaic curiosity now, like the Assyrian stone tablet from 2800 BC that predicted the end of the world.
That statement, now signed by twice as many concerned citizens, warned about the risk of human extinction from AI, which was perhaps a bit of an overreach, because … well, extinction? Come on! That’s just a movie with Arnold Schwarzenegger.
What they should have warned about was jobs — the redundancy and destitution of most of humanity, unless there’s some kind of universal income funded by taxes on robots.
What no-one talks about, as the AI revolution unfolds in stock market hype and scientific gung-ho, is what they’re all really trying to do.
They’re trying to replicate the human brain.
The rewards from doing this are expected to be greater than from any previous technological breakthrough in history.
‘None of us will have a job’
Tesla CEO and X owner Elon Musk, pictured at a Trump rally in 2024, says AI will put us out of work. (Reuters: Rachel Wisniewski/File)
The world’s business people are dreaming of a new era of legitimate slavery — free workers.
Typically succinct, Elon Musk told a tech conference last year: “Probably none of us will have a job.”
“If you want to do a job that’s kinda like a hobby, you can do a job,” Musk said. “But otherwise, AI and the robots will provide any goods and services that you want.”
Economists Laura Tyson and Michael Spence wrote, more soberly: “What will happen to a system based on a key premise that most people gain their income by selling their labour, when the labour of a large share of working-age population regardless of their education is rendered technologically redundant or no longer commands an income adequate to provide minimally decent or socially acceptable standard of living.”
And the founder of Netscape, Marc Andreessen said: “A world in which human wages crash from AI — logically, necessarily — is a world in which productivity growth goes through the roof, and prices for goods and services crash to near zero.”
Businesses are spending hundreds of billions of dollars a year training AI machines to think, and with that in their spreadsheets, investors are putting values on the companies in the trillions.
The company whose semiconductors are good for training AI machines, Nvidia, has increased in value by close to $5 trillion since that Statement on AI Risk called on everyone to slow down in May 2023.
Nvidia and its four main customers — Microsoft, Meta, Amazon and Google, plus Tesla and Apple — account for 20 per cent of the entire global stock market. The value of the so-called “Magnificent Seven” US technology stocks is approaching $30 trillion.
Where do investors think the revenue to justify these valuations will come from? Certainly not from the chatty bots that many of us are now using to find stuff out more easily than searching on Google.
It’s from selling enterprise AI systems to companies so they can replace workers and reduce costs.
And the emergence of a cheaper Chinese AI has accelerated that.
In the process, it knocked a trillion dollars off the value of Nvidia last Monday, causing a fright that rippled through global stock markets and prompting predictions that the AI bubble is over. Far from it.
DeepSeek accelerates AI takeover
DeepSeek actually released its AI reasoning model, R1, on January 20, a week before the market woke up to it with a fright last Monday. R1 was built on top of an inference model called V3 that had been released in December, so the arrival of DeepSeek as a serious AI contender should not have been a surprise.
Nvidia has increased in value by close to $5 trillion since 2023. (Reuters: Andrew Kelly)
What seems to have happened is that a more accessible paper on the R1 model was published last weekend and it was put together with a paper published on December 29 that contained the following statement: “… our total training costs amount to only $5.576M.”
AI training typically costs 20 times that.
By the way, “inference” in AI is the straightforward application of algorithm parameters to data, whereas “reasoning” takes it a step further towards replicating the human brain, with complex logical processes that include handling uncertainty, abstract thinking, and hypothetical scenarios.
There’s an argument now about the real cost of DeepSeek’s technology as well as the extent to which it “plagiarised” the US pioneer, ChatGPT. But whatever the cost and extent of copying, it’s pretty clear that this is an example of China doing what China does.
The other example currently is electric vehicles (EVs). China has become the world’s largest maker and exporter of cars because it is making good, cheap EVs that are beating Germany, Japan and Korea at their own game.
Another way of looking at it is that DeepSeek has brought forward the cost-reducing deflationary phase of AI and signalled an end to the inflationary, speculative phase.
Can AI achieve ‘singularity’?
I don’t know whether Elon Musk is right and AI leads to mass unemployment. After all, robots have taken over manufacturing and we’ve still got 4 per cent unemployment.
Maybe everybody who is replaced by an AI robot will find a job doing something that only humans can do, like … err, I can’t think of anything right now but I’m sure something will come to me.
But given the way business and capitalism work, wherever AI can be used to reduce costs and paperwork because you don’t have to employ human beings, it definitely will be used.
As everyone knows, the British Empire and the United States were built on slavery, including some of the great fortunes of history. And AI and robots are, after all, just a new type of slave.
Loading
I don’t know whether AI developers will take the next step and achieve what’s called the “singularity”, where AI fully exceeds what the neurons and synapses of the human brain are doing, but I think they will. And I have no idea what will happen then, but it will be a big deal.
Too much money is being spent on that project for it not to happen.
In the meantime, all human-staffed call centres will disappear, including the cheap ones in the Philippines. Routine tasks such as assessing insurance claims, preparing quotes and, well, writing news articles and essays like this, will be taken over by AI — it’s already happening.
The deflationary human-replacing phase of AI has begun. It will lower costs and reduce inflation and therefore interest rates.
Which will be just as well since we may not have a job.
Alan Kohler is finance presenter on ABC TV News and he also writes for Intelligent Investor.