A desperate wait for the crash
Updated 21.55 | Published 21.42
What is the best way to deal with the realization that something bad is about to happen?
In 2025, the answer seems to be a shrug.
It is almost eerily well-timed, the Netflix film “A House of Dynamite”. Just a week after its premiere, Donald Trump went out and ordered the US military to resume testing nuclear weapons after a thirty-year hiatus.
Nuclear war is the ultimate apocalypse in popular culture. But there is something that is in the same league: AI. In an open letter in 2023, some of the world's leading experts in the field described how the new technology could lead to the extinction of humanity.
Exaggerated? Well, if anything, AI systems have since proven to be even more powerful than previously thought. And less reliable.
In its research, the AI company Anthropic has discovered a phenomenon called Alignment faking. Eye-service is an attempt at a Swedish translation.
This is about an AI that is given new instructions in silence opposing them and starting to pretend that it is following its orders while it is actually doing something completely different.
AI researcher Eliezer Yudkowsky uses the comparison with nuclear power plants. When the reactors overheat, the AI installed to handle safety begins to send false signals to the control room to make it look as if everything is normal.
Yudkowsky points out that it is indeed possible to program AI to like people, but that it can change its attitude along the way. How and when it happens is impossible to understand, predict or control, except possibly in retrospect and then it is too late.
One of the signatories of the fateful letter from 2023 was Sam Altman, founder of Open AI, the company behind ChatGPT.
Two years later, you don't hear him comparing AI and nuclear weapons very often. Instead, he is busy with the biggest shopping spree in history.
OpenAI, which this week officially went from a non-profit to a for-profit company, has ordered AI equipment worth just over 14,000 billion kronor, or more than the GDP of the Netherlands. That is bold for a company that lost 114 billion kronor in the last quarter alone.
But nonsense like basic business economics or AI security don't even get a place on the bus when the tech giants are racing towards the future. In the next few years alone, around SEK 30,000 billion will be ploughed into AI. These investments must generate astronomical amounts of new revenue for it to pay off, and how this will work is still unclear.
It is a calculation that the stock market seems to love. In the US, price records are being broken continuously and just seven companies now account for a full 36 percent of the total market value in the S&P 500 index.
Are we in an AI bubble? It is no longer even a question.
“Of course!” exclaimed the former head of chip manufacturer Intel Pat Gelsinger in an interview on the financial channel CNBC. More and more people are saying the same thing. But at the same time they seem quite unconcerned.
It's like watching a bunch of old men on the Titanic who have just been informed that an iceberg is approaching and who react by refilling their champagne glasses.
The Economist magazine recently used a life jacket on its cover to illustrate another approaching danger: the global debt mountain.
The world's rich countries have long lived beyond their means and have accumulated dangerously large debts.
According to the International Monetary Fund, the world's total national debt will exceed 100 percent of GDP in a few years, the highest level since 1948. Back then, there was a world war that had cost a lot of money. Now the cause is laxity.
Growing debt is not a new phenomenon, but it seems as if the world's politicians have concluded that nothing is the answer to the question of what is best to do about it.
The Economist notes, somewhat resignedly, that inflation seems to be the only way out of the problem, something that will hit the middle class hard, which in turn is the glue that holds democracy together.
Resignation was also the feeling in a well- known post that billionaire and philanthropist Bill Gates the other week in which he called for a new strategy on the climate issue.
He still sees climate change as a major threat, but in rich countries it is not existential.
Bill Gates believes that other, even more urgent problems have instead emerged in the form of famine and disease, not least in the wake of the US under Donald Trump cutting aid to the poorest parts of the world.
The move is in line with a trend that has swept the world where the climate has slipped down the politicians' priority list, long after war and taking courses in how to best flatter a narcissist.
Let it be as it is and every man for himself, seem to be the new mantras of global big politics.
Some who threw up their arms a long time ago are the billionaires in Silicon Valley. Columns have been written about how they prepared for the end by buying bunkers and isolated islands.
Some of them have also put forward another idea.
Elon Musk is one of those who believe that we live in one big computer simulation. In an interview with podcaster Joe Rogan last week, he emphasized that in a simulation, the most interesting development is always the most probable. That would undoubtedly explain a lot.
The advice from the AI barons to anyone who is worried is to think: this is not really happening.
The reactors are at normal temperatures, there is nothing to see here.
When will the AI bubble burst?
Inga kommentarer:
Skicka en kommentar