Here is an AI skeptic, and he is maybe the biggest Economics big-shot.
MIT Technology Review reports:
A Nobel laureate on the economics of artificial intelligence
Daron Acemoglu has long studied technology-driven growth. He thinks we should slow down and make sure we’re using AI the right way.
For all the talk about artificial intelligence upending the world, its economic effects remain uncertain. But Institute Professor and 2024 Nobel winner Daron Acemoglu has some insights.
Despite some predictions that AI will double US GDP growth, Acemoglu expects it to increase GDP by 1.1% to 1.6% over the next 10 years, with a roughly 0.05% annual gain in productivity.
He is widely praised for his analysis of big trends, but I have criticized him
here and
here.
I tried asking an AI LLM how I could give a politically respectable account of how nations succeed and fail. It said to give all the credit
to institutions, and occasionally blame colonialism, but never draw attention to human capital. In other words, say exactly what Acemoglu says.
He pretends to be data-driven, but I do not believe it. He just gets ahead by telling people what they want to hear.
I am reminded of Economist Paul Krugman writing that the internet's effect on the economy would be no greater than the fax machine's.
I occasionally see others argue that AI is not intelligent, or is still looking for the killer app, or is just a gimmick,
or has not progressed. For example, Gary Marcus says AI progress has ended.
Wake up. AI has made very consistent progress over the last 15 years, and has shown no sign of slowing down. Its
economic value is already many billions, particularly in image processing, text drafting, and software coding.
OpenAI announced that it will buy Jony Ivy's startup for $6.5B, and Windsurf for $3B. And it intends to lose
$44B before it makes a profit. It is supposed to be a nonprofit, so it still has to legally figure out how
to be a for-profit. So there is legitimate skepticism about whether its plans will pay off.
Maybe not, but ChatGPT technology is incredibly useful, and the latest models from Google, Anthropic, Meta,
and the Chinese are getting really good.
The wait for self-driving cars and humanoid robots is frustrating.
A few decades ago, the whole world was computerizing, and economists
complained that there were no measurable productivity gains. Acemoglu even wrote a 2014 paper complaining about it. So why were companies computerizing?
The computers were leading to great efficiencies that the economists were not measuring.
We are going to see the same thing with AI. Soon the whole world will be running on AI, and
it will work so well that we will wonder how we got by without it. But Acemoglu and other
economists will deny that AI has any economic impact.
Update: New interview:
Nobel Laureate Busts the AI Hype
MIT Sloan Management Review
While many people are predicting that AI will rapidly transform the economy, MIT economist Daron Acemoglu offers a more modest forecast. Drawing from his research, in this brief video he explains why AI might automate only 5% of tasks and add just 1% to global GDP over the next decade. Acemoglu also asserts that AI’s potential is less clear than the internet’s was and that human judgment trumps algorithms, and he challenges leaders to innovate with AI rather than simply slash costs.
He doubles down on his argument that AI is over-hyped, and will be of little consequence.
He says that he cannot be replaced by AI, but I think he can.